[GitHub/mt8127/android_kernel_alcatel_ttab.git] / block / blk-ioc.c

/*
 * Functions related to io context handling
 */
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/bootmem.h>	/* for max_pfn/max_low_pfn */
#include <linux/slab.h>

#include "blk.h"

/*
 * For io context allocations
 */
static struct kmem_cache *iocontext_cachep;

/**
 * get_io_context - increment reference count to io_context
 * @ioc: io_context to get
 *
 * Increment reference count to @ioc.
 */
void get_io_context(struct io_context *ioc)
{
	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
	atomic_long_inc(&ioc->refcount);
}
EXPORT_SYMBOL(get_io_context);

static void icq_free_icq_rcu(struct rcu_head *head)
{
	struct io_cq *icq = container_of(head, struct io_cq, __rcu_head);

	kmem_cache_free(icq->__rcu_icq_cache, icq);
}

/* Exit an icq. Called with both ioc and q locked. */
static void ioc_exit_icq(struct io_cq *icq)
{
	struct elevator_type *et = icq->q->elevator->type;

	if (icq->flags & ICQ_EXITED)
		return;

	if (et->ops.elevator_exit_icq_fn)
		et->ops.elevator_exit_icq_fn(icq);

	icq->flags |= ICQ_EXITED;
}

/* Release an icq.  Called with both ioc and q locked. */
static void ioc_destroy_icq(struct io_cq *icq)
{
	struct io_context *ioc = icq->ioc;
	struct request_queue *q = icq->q;
	struct elevator_type *et = q->elevator->type;

	lockdep_assert_held(&ioc->lock);
	lockdep_assert_held(q->queue_lock);

	radix_tree_delete(&ioc->icq_tree, icq->q->id);
	hlist_del_init(&icq->ioc_node);
	list_del_init(&icq->q_node);

	/*
	 * Both setting lookup hint to and clearing it from @icq are done
	 * under queue_lock.  If it's not pointing to @icq now, it never
	 * will.  Hint assignment itself can race safely.
	 */
	if (rcu_dereference_raw(ioc->icq_hint) == icq)
		rcu_assign_pointer(ioc->icq_hint, NULL);

	ioc_exit_icq(icq);

	/*
	 * @icq->q might have gone away by the time RCU callback runs
	 * making it impossible to determine icq_cache.  Record it in @icq.
	 */
	icq->__rcu_icq_cache = et->icq_cache;
	call_rcu(&icq->__rcu_head, icq_free_icq_rcu);
}

/*
 * Slow path for ioc release in put_io_context().  Performs double-lock
 * dancing to unlink all icq's and then frees ioc.
 */
static void ioc_release_fn(struct work_struct *work)
{
	struct io_context *ioc = container_of(work, struct io_context,
					      release_work);
	unsigned long flags;

	/*
	 * Exiting icq may call into put_io_context() through elevator
	 * which will trigger lockdep warning.  The ioc's are guaranteed to
	 * be different, use a different locking subclass here.  Use
	 * irqsave variant as there's no spin_lock_irq_nested().
	 */
	spin_lock_irqsave_nested(&ioc->lock, flags, 1);

	while (!hlist_empty(&ioc->icq_list)) {
		struct io_cq *icq = hlist_entry(ioc->icq_list.first,
						struct io_cq, ioc_node);
		struct request_queue *q = icq->q;

		if (spin_trylock(q->queue_lock)) {
			ioc_destroy_icq(icq);
			spin_unlock(q->queue_lock);
		} else {
			spin_unlock_irqrestore(&ioc->lock, flags);
			cpu_relax();
			spin_lock_irqsave_nested(&ioc->lock, flags, 1);
		}
	}

	spin_unlock_irqrestore(&ioc->lock, flags);

	kmem_cache_free(iocontext_cachep, ioc);
}

/**
 * put_io_context - put a reference of io_context
 * @ioc: io_context to put
 *
 * Decrement reference count of @ioc and release it if the count reaches
 * zero.
 */
void put_io_context(struct io_context *ioc)
{
	unsigned long flags;
	bool free_ioc = false;

	if (ioc == NULL)
		return;

	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);

	/*
	 * Releasing ioc requires reverse order double locking and we may
	 * already be holding a queue_lock.  Do it asynchronously from wq.
	 */
	if (atomic_long_dec_and_test(&ioc->refcount)) {
		spin_lock_irqsave(&ioc->lock, flags);
		if (!hlist_empty(&ioc->icq_list))
			schedule_work(&ioc->release_work);
		else
			free_ioc = true;
		spin_unlock_irqrestore(&ioc->lock, flags);
	}

	if (free_ioc)
		kmem_cache_free(iocontext_cachep, ioc);
}
EXPORT_SYMBOL(put_io_context);

/**
 * put_io_context_active - put active reference on ioc
 * @ioc: ioc of interest
 *
 * Undo get_io_context_active().  If active reference reaches zero after
 * put, @ioc can never issue further IOs and ioscheds are notified.
 */
void put_io_context_active(struct io_context *ioc)
{
	unsigned long flags;
	struct io_cq *icq;

	if (!atomic_dec_and_test(&ioc->active_ref)) {
		put_io_context(ioc);
		return;
	}

	/*
	 * Need ioc lock to walk icq_list and q lock to exit icq.  Perform
	 * reverse double locking.  Read comment in ioc_release_fn() for
	 * explanation on the nested locking annotation.
	 */
retry:
	spin_lock_irqsave_nested(&ioc->lock, flags, 1);
	hlist_for_each_entry(icq, &ioc->icq_list, ioc_node) {
		if (icq->flags & ICQ_EXITED)
			continue;
		if (spin_trylock(icq->q->queue_lock)) {
			ioc_exit_icq(icq);
			spin_unlock(icq->q->queue_lock);
		} else {
			spin_unlock_irqrestore(&ioc->lock, flags);
			cpu_relax();
			goto retry;
		}
	}
	spin_unlock_irqrestore(&ioc->lock, flags);

	put_io_context(ioc);
}

/* Called by the exiting task */
void exit_io_context(struct task_struct *task)
{
	struct io_context *ioc;

	task_lock(task);
	ioc = task->io_context;
	task->io_context = NULL;
	task_unlock(task);

	atomic_dec(&ioc->nr_tasks);
	put_io_context_active(ioc);
}

/**
 * ioc_clear_queue - break any ioc association with the specified queue
 * @q: request_queue being cleared
 *
 * Walk @q->icq_list and exit all io_cq's.  Must be called with @q locked.
 */
void ioc_clear_queue(struct request_queue *q)
{
	lockdep_assert_held(q->queue_lock);

	while (!list_empty(&q->icq_list)) {
		struct io_cq *icq = list_entry(q->icq_list.next,
					       struct io_cq, q_node);
		struct io_context *ioc = icq->ioc;

		spin_lock(&ioc->lock);
		ioc_destroy_icq(icq);
		spin_unlock(&ioc->lock);
	}
}

int create_task_io_context(struct task_struct *task, gfp_t gfp_flags, int node)
{
	struct io_context *ioc;
	int ret;

	ioc = kmem_cache_alloc_node(iocontext_cachep, gfp_flags | __GFP_ZERO,
				    node);
	if (unlikely(!ioc))
		return -ENOMEM;

	/* initialize */
	atomic_long_set(&ioc->refcount, 1);
	atomic_set(&ioc->nr_tasks, 1);
	atomic_set(&ioc->active_ref, 1);
	spin_lock_init(&ioc->lock);
	INIT_RADIX_TREE(&ioc->icq_tree, GFP_ATOMIC | __GFP_HIGH);
	INIT_HLIST_HEAD(&ioc->icq_list);
	INIT_WORK(&ioc->release_work, ioc_release_fn);

	/*
	 * Try to install.  ioc shouldn't be installed if someone else
	 * already did or @task, which isn't %current, is exiting.  Note
	 * that we need to allow ioc creation on exiting %current as exit
	 * path may issue IOs from e.g. exit_files().  The exit path is
	 * responsible for not issuing IO after exit_io_context().
	 */
	task_lock(task);
	if (!task->io_context &&
	    (task == current || !(task->flags & PF_EXITING)))
		task->io_context = ioc;
	else
		kmem_cache_free(iocontext_cachep, ioc);

	ret = task->io_context ? 0 : -EBUSY;

	task_unlock(task);

	return ret;
}

/**
 * get_task_io_context - get io_context of a task
 * @task: task of interest
 * @gfp_flags: allocation flags, used if allocation is necessary
 * @node: allocation node, used if allocation is necessary
 *
 * Return io_context of @task.  If it doesn't exist, it is created with
 * @gfp_flags and @node.  The returned io_context has its reference count
 * incremented.
 *
 * This function always goes through task_lock() and it's better to use
 * %current->io_context + get_io_context() for %current.
 */
struct io_context *get_task_io_context(struct task_struct *task,
				       gfp_t gfp_flags, int node)
{
	struct io_context *ioc;

	might_sleep_if(gfp_flags & __GFP_WAIT);

	do {
		task_lock(task);
		ioc = task->io_context;
		if (likely(ioc)) {
			get_io_context(ioc);
			task_unlock(task);
			return ioc;
		}
		task_unlock(task);
	} while (!create_task_io_context(task, gfp_flags, node));

	return NULL;
}
EXPORT_SYMBOL(get_task_io_context);

/**
 * ioc_lookup_icq - lookup io_cq from ioc
 * @ioc: the associated io_context
 * @q: the associated request_queue
 *
 * Look up io_cq associated with @ioc - @q pair from @ioc.  Must be called
 * with @q->queue_lock held.
 */
struct io_cq *ioc_lookup_icq(struct io_context *ioc, struct request_queue *q)
{
	struct io_cq *icq;

	lockdep_assert_held(q->queue_lock);

	/*
	 * icq's are indexed from @ioc using radix tree and hint pointer,
	 * both of which are protected with RCU.  All removals are done
	 * holding both q and ioc locks, and we're holding q lock - if we
	 * find a icq which points to us, it's guaranteed to be valid.
	 */
	rcu_read_lock();
	icq = rcu_dereference(ioc->icq_hint);
	if (icq && icq->q == q)
		goto out;

	icq = radix_tree_lookup(&ioc->icq_tree, q->id);
	if (icq && icq->q == q)
		rcu_assign_pointer(ioc->icq_hint, icq);	/* allowed to race */
	else
		icq = NULL;
out:
	rcu_read_unlock();
	return icq;
}
EXPORT_SYMBOL(ioc_lookup_icq);

/**
 * ioc_create_icq - create and link io_cq
 * @ioc: io_context of interest
 * @q: request_queue of interest
 * @gfp_mask: allocation mask
 *
 * Make sure io_cq linking @ioc and @q exists.  If icq doesn't exist, they
 * will be created using @gfp_mask.
 *
 * The caller is responsible for ensuring @ioc won't go away and @q is
 * alive and will stay alive until this function returns.
 */
struct io_cq *ioc_create_icq(struct io_context *ioc, struct request_queue *q,
			     gfp_t gfp_mask)
{
	struct elevator_type *et = q->elevator->type;
	struct io_cq *icq;

	/* allocate stuff */
	icq = kmem_cache_alloc_node(et->icq_cache, gfp_mask | __GFP_ZERO,
				    q->node);
	if (!icq)
		return NULL;

	if (radix_tree_preload(gfp_mask) < 0) {
		kmem_cache_free(et->icq_cache, icq);
		return NULL;
	}

	icq->ioc = ioc;
	icq->q = q;
	INIT_LIST_HEAD(&icq->q_node);
	INIT_HLIST_NODE(&icq->ioc_node);

	/* lock both q and ioc and try to link @icq */
	spin_lock_irq(q->queue_lock);
	spin_lock(&ioc->lock);

	if (likely(!radix_tree_insert(&ioc->icq_tree, q->id, icq))) {
		hlist_add_head(&icq->ioc_node, &ioc->icq_list);
		list_add(&icq->q_node, &q->icq_list);
		if (et->ops.elevator_init_icq_fn)
			et->ops.elevator_init_icq_fn(icq);
	} else {
		kmem_cache_free(et->icq_cache, icq);
		icq = ioc_lookup_icq(ioc, q);
		if (!icq)
			printk(KERN_ERR "cfq: icq link failed!\n");
	}

	spin_unlock(&ioc->lock);
	spin_unlock_irq(q->queue_lock);
	radix_tree_preload_end();
	return icq;
}

static int __init blk_ioc_init(void)
{
	iocontext_cachep = kmem_cache_create("blkdev_ioc",
			sizeof(struct io_context), 0, SLAB_PANIC, NULL);
	return 0;
}
subsys_initcall(blk_ioc_init);
Commit	Line	Data
86db1e29 JA	1	/*
	2	* Functions related to io context handling
	3	*/
	4	#include <linux/kernel.h>
	5	#include <linux/module.h>
	6	#include <linux/init.h>
	7	#include <linux/bio.h>
	8	#include <linux/blkdev.h>
	9	#include <linux/bootmem.h> /* for max_pfn/max_low_pfn */
5a0e3ad6	10	#include <linux/slab.h>
86db1e29 JA	11
	12	#include "blk.h"
	13
	14	/*
	15	* For io context allocations
	16	*/
	17	static struct kmem_cache *iocontext_cachep;
	18
6e736be7 TH	19	/**
	20	* get_io_context - increment reference count to io_context
	21	* @ioc: io_context to get
	22	*
	23	* Increment reference count to @ioc.
	24	*/
	25	void get_io_context(struct io_context *ioc)
	26	{
	27	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
	28	atomic_long_inc(&ioc->refcount);
	29	}
	30	EXPORT_SYMBOL(get_io_context);
	31
7e5a8794 TH	32	static void icq_free_icq_rcu(struct rcu_head *head)
	33	{
	34	struct io_cq *icq = container_of(head, struct io_cq, __rcu_head);
	35
	36	kmem_cache_free(icq->__rcu_icq_cache, icq);
	37	}
	38
621032ad	39	/* Exit an icq. Called with both ioc and q locked. */
7e5a8794	40	static void ioc_exit_icq(struct io_cq *icq)
621032ad TH	41	{
	42	struct elevator_type *et = icq->q->elevator->type;
	43
	44	if (icq->flags & ICQ_EXITED)
	45	return;
	46
	47	if (et->ops.elevator_exit_icq_fn)
	48	et->ops.elevator_exit_icq_fn(icq);
	49
	50	icq->flags \|= ICQ_EXITED;
	51	}
	52
	53	/* Release an icq. Called with both ioc and q locked. */
	54	static void ioc_destroy_icq(struct io_cq *icq)
7e5a8794 TH	55	{
	56	struct io_context *ioc = icq->ioc;
	57	struct request_queue *q = icq->q;
	58	struct elevator_type *et = q->elevator->type;
	59
	60	lockdep_assert_held(&ioc->lock);
	61	lockdep_assert_held(q->queue_lock);
	62
	63	radix_tree_delete(&ioc->icq_tree, icq->q->id);
	64	hlist_del_init(&icq->ioc_node);
	65	list_del_init(&icq->q_node);
	66
	67	/*
	68	* Both setting lookup hint to and clearing it from @icq are done
	69	* under queue_lock. If it's not pointing to @icq now, it never
	70	* will. Hint assignment itself can race safely.
	71	*/
	72	if (rcu_dereference_raw(ioc->icq_hint) == icq)
	73	rcu_assign_pointer(ioc->icq_hint, NULL);
	74
621032ad	75	ioc_exit_icq(icq);
7e5a8794 TH	76
	77	/*
	78	* @icq->q might have gone away by the time RCU callback runs
	79	* making it impossible to determine icq_cache. Record it in @icq.
	80	*/
	81	icq->__rcu_icq_cache = et->icq_cache;
	82	call_rcu(&icq->__rcu_head, icq_free_icq_rcu);
	83	}
	84
b2efa052 TH	85	/*
b2efa052 TH	86	* Slow path for ioc release in put_io_context(). Performs double-lock
c5869807	87	* dancing to unlink all icq's and then frees ioc.
b2efa052 TH	88	*/
b2efa052 TH	89	static void ioc_release_fn(struct work_struct *work)
86db1e29	90	{
b2efa052 TH	91	struct io_context *ioc = container_of(work, struct io_context,
b2efa052 TH	92	release_work);
d8c66c5d	93	unsigned long flags;
b2efa052	94
d8c66c5d TH	95	/*
	96	* Exiting icq may call into put_io_context() through elevator
	97	* which will trigger lockdep warning. The ioc's are guaranteed to
	98	* be different, use a different locking subclass here. Use
	99	* irqsave variant as there's no spin_lock_irq_nested().
	100	*/
	101	spin_lock_irqsave_nested(&ioc->lock, flags, 1);
b2efa052	102
c5869807 TH	103	while (!hlist_empty(&ioc->icq_list)) {
	104	struct io_cq *icq = hlist_entry(ioc->icq_list.first,
	105	struct io_cq, ioc_node);
2274b029 TH	106	struct request_queue *q = icq->q;
	107
	108	if (spin_trylock(q->queue_lock)) {
621032ad	109	ioc_destroy_icq(icq);
2274b029 TH	110	spin_unlock(q->queue_lock);
	111	} else {
	112	spin_unlock_irqrestore(&ioc->lock, flags);
	113	cpu_relax();
	114	spin_lock_irqsave_nested(&ioc->lock, flags, 1);
b2efa052	115	}
b2efa052	116	}
ffc4e759	117
2274b029	118	spin_unlock_irqrestore(&ioc->lock, flags);
b2efa052 TH	119
b2efa052 TH	120	kmem_cache_free(iocontext_cachep, ioc);
86db1e29 JA	121	}
86db1e29 JA	122
42ec57a8 TH	123	/**
	124	* put_io_context - put a reference of io_context
	125	* @ioc: io_context to put
	126	*
	127	* Decrement reference count of @ioc and release it if the count reaches
11a3122f	128	* zero.
86db1e29	129	*/
11a3122f	130	void put_io_context(struct io_context *ioc)
86db1e29	131	{
b2efa052	132	unsigned long flags;
ff8c1474	133	bool free_ioc = false;
b2efa052	134
86db1e29	135	if (ioc == NULL)
42ec57a8	136	return;
86db1e29	137
42ec57a8	138	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
86db1e29	139
b2efa052	140	/*
11a3122f TH	141	* Releasing ioc requires reverse order double locking and we may
11a3122f TH	142	* already be holding a queue_lock. Do it asynchronously from wq.
b2efa052	143	*/
11a3122f TH	144	if (atomic_long_dec_and_test(&ioc->refcount)) {
	145	spin_lock_irqsave(&ioc->lock, flags);
	146	if (!hlist_empty(&ioc->icq_list))
	147	schedule_work(&ioc->release_work);
ff8c1474 XF	148	else
ff8c1474 XF	149	free_ioc = true;
11a3122f	150	spin_unlock_irqrestore(&ioc->lock, flags);
b2efa052	151	}
ff8c1474 XF	152
	153	if (free_ioc)
	154	kmem_cache_free(iocontext_cachep, ioc);
86db1e29	155	}
b2efa052	156	EXPORT_SYMBOL(put_io_context);
86db1e29	157
f6e8d01b TH	158	/**
	159	* put_io_context_active - put active reference on ioc
	160	* @ioc: ioc of interest
	161	*
	162	* Undo get_io_context_active(). If active reference reaches zero after
	163	* put, @ioc can never issue further IOs and ioscheds are notified.
	164	*/
	165	void put_io_context_active(struct io_context *ioc)
86db1e29	166	{
621032ad	167	unsigned long flags;
f6e8d01b	168	struct io_cq *icq;
86db1e29	169
f6e8d01b	170	if (!atomic_dec_and_test(&ioc->active_ref)) {
621032ad TH	171	put_io_context(ioc);
	172	return;
	173	}
	174
	175	/*
	176	* Need ioc lock to walk icq_list and q lock to exit icq. Perform
	177	* reverse double locking. Read comment in ioc_release_fn() for
	178	* explanation on the nested locking annotation.
	179	*/
	180	retry:
	181	spin_lock_irqsave_nested(&ioc->lock, flags, 1);
b67bfe0d	182	hlist_for_each_entry(icq, &ioc->icq_list, ioc_node) {
621032ad TH	183	if (icq->flags & ICQ_EXITED)
	184	continue;
	185	if (spin_trylock(icq->q->queue_lock)) {
	186	ioc_exit_icq(icq);
	187	spin_unlock(icq->q->queue_lock);
	188	} else {
	189	spin_unlock_irqrestore(&ioc->lock, flags);
	190	cpu_relax();
	191	goto retry;
	192	}
	193	}
	194	spin_unlock_irqrestore(&ioc->lock, flags);
	195
11a3122f	196	put_io_context(ioc);
86db1e29 JA	197	}
86db1e29 JA	198
f6e8d01b TH	199	/* Called by the exiting task */
	200	void exit_io_context(struct task_struct *task)
	201	{
	202	struct io_context *ioc;
	203
	204	task_lock(task);
	205	ioc = task->io_context;
	206	task->io_context = NULL;
	207	task_unlock(task);
	208
	209	atomic_dec(&ioc->nr_tasks);
	210	put_io_context_active(ioc);
	211	}
	212
7e5a8794 TH	213	/**
	214	* ioc_clear_queue - break any ioc association with the specified queue
	215	* @q: request_queue being cleared
	216	*
	217	* Walk @q->icq_list and exit all io_cq's. Must be called with @q locked.
	218	*/
	219	void ioc_clear_queue(struct request_queue *q)
	220	{
	221	lockdep_assert_held(q->queue_lock);
	222
	223	while (!list_empty(&q->icq_list)) {
	224	struct io_cq *icq = list_entry(q->icq_list.next,
	225	struct io_cq, q_node);
	226	struct io_context *ioc = icq->ioc;
	227
	228	spin_lock(&ioc->lock);
621032ad	229	ioc_destroy_icq(icq);
7e5a8794 TH	230	spin_unlock(&ioc->lock);
	231	}
	232	}
	233
24acfc34	234	int create_task_io_context(struct task_struct *task, gfp_t gfp_flags, int node)
86db1e29	235	{
df415656	236	struct io_context *ioc;
3c9c708c	237	int ret;
86db1e29	238
42ec57a8 TH	239	ioc = kmem_cache_alloc_node(iocontext_cachep, gfp_flags \| __GFP_ZERO,
	240	node);
	241	if (unlikely(!ioc))
24acfc34	242	return -ENOMEM;
42ec57a8 TH	243
	244	/* initialize */
	245	atomic_long_set(&ioc->refcount, 1);
4638a83e	246	atomic_set(&ioc->nr_tasks, 1);
f6e8d01b	247	atomic_set(&ioc->active_ref, 1);
42ec57a8	248	spin_lock_init(&ioc->lock);
c5869807 TH	249	INIT_RADIX_TREE(&ioc->icq_tree, GFP_ATOMIC \| __GFP_HIGH);
c5869807 TH	250	INIT_HLIST_HEAD(&ioc->icq_list);
b2efa052	251	INIT_WORK(&ioc->release_work, ioc_release_fn);
86db1e29	252
fd638368 TH	253	/*
	254	* Try to install. ioc shouldn't be installed if someone else
	255	* already did or @task, which isn't %current, is exiting. Note
	256	* that we need to allow ioc creation on exiting %current as exit
	257	* path may issue IOs from e.g. exit_files(). The exit path is
	258	* responsible for not issuing IO after exit_io_context().
	259	*/
6e736be7	260	task_lock(task);
fd638368 TH	261	if (!task->io_context &&
fd638368 TH	262	(task == current \|\| !(task->flags & PF_EXITING)))
6e736be7	263	task->io_context = ioc;
f2dbd76a	264	else
6e736be7	265	kmem_cache_free(iocontext_cachep, ioc);
3c9c708c ED	266
	267	ret = task->io_context ? 0 : -EBUSY;
	268
6e736be7	269	task_unlock(task);
24acfc34	270
3c9c708c	271	return ret;
86db1e29	272	}
86db1e29	273
6e736be7 TH	274	/**
	275	* get_task_io_context - get io_context of a task
	276	* @task: task of interest
	277	* @gfp_flags: allocation flags, used if allocation is necessary
	278	* @node: allocation node, used if allocation is necessary
	279	*
	280	* Return io_context of @task. If it doesn't exist, it is created with
	281	* @gfp_flags and @node. The returned io_context has its reference count
	282	* incremented.
86db1e29	283	*
6e736be7	284	* This function always goes through task_lock() and it's better to use
f2dbd76a	285	* %current->io_context + get_io_context() for %current.
86db1e29	286	*/
6e736be7 TH	287	struct io_context get_task_io_context(struct task_struct task,
6e736be7 TH	288	gfp_t gfp_flags, int node)
86db1e29	289	{
6e736be7	290	struct io_context *ioc;
86db1e29	291
6e736be7 TH	292	might_sleep_if(gfp_flags & __GFP_WAIT);
6e736be7 TH	293
f2dbd76a TH	294	do {
	295	task_lock(task);
	296	ioc = task->io_context;
	297	if (likely(ioc)) {
	298	get_io_context(ioc);
	299	task_unlock(task);
	300	return ioc;
	301	}
6e736be7	302	task_unlock(task);
24acfc34	303	} while (!create_task_io_context(task, gfp_flags, node));
6e736be7	304
f2dbd76a	305	return NULL;
86db1e29	306	}
6e736be7	307	EXPORT_SYMBOL(get_task_io_context);
86db1e29	308
47fdd4ca TH	309	/**
	310	* ioc_lookup_icq - lookup io_cq from ioc
	311	* @ioc: the associated io_context
	312	* @q: the associated request_queue
	313	*
	314	* Look up io_cq associated with @ioc - @q pair from @ioc. Must be called
	315	* with @q->queue_lock held.
	316	*/
	317	struct io_cq ioc_lookup_icq(struct io_context ioc, struct request_queue *q)
	318	{
	319	struct io_cq *icq;
	320
	321	lockdep_assert_held(q->queue_lock);
	322
	323	/*
	324	* icq's are indexed from @ioc using radix tree and hint pointer,
	325	* both of which are protected with RCU. All removals are done
	326	* holding both q and ioc locks, and we're holding q lock - if we
	327	* find a icq which points to us, it's guaranteed to be valid.
	328	*/
	329	rcu_read_lock();
	330	icq = rcu_dereference(ioc->icq_hint);
	331	if (icq && icq->q == q)
	332	goto out;
	333
	334	icq = radix_tree_lookup(&ioc->icq_tree, q->id);
	335	if (icq && icq->q == q)
	336	rcu_assign_pointer(ioc->icq_hint, icq); /* allowed to race */
	337	else
	338	icq = NULL;
	339	out:
	340	rcu_read_unlock();
	341	return icq;
	342	}
	343	EXPORT_SYMBOL(ioc_lookup_icq);
	344
f1f8cc94 TH	345	/**
f1f8cc94 TH	346	* ioc_create_icq - create and link io_cq
24acfc34	347	* @ioc: io_context of interest
f1f8cc94 TH	348	* @q: request_queue of interest
	349	* @gfp_mask: allocation mask
	350	*
24acfc34 TH	351	* Make sure io_cq linking @ioc and @q exists. If icq doesn't exist, they
24acfc34 TH	352	* will be created using @gfp_mask.
f1f8cc94 TH	353	*
	354	* The caller is responsible for ensuring @ioc won't go away and @q is
	355	* alive and will stay alive until this function returns.
	356	*/
24acfc34 TH	357	struct io_cq ioc_create_icq(struct io_context ioc, struct request_queue *q,
24acfc34 TH	358	gfp_t gfp_mask)
f1f8cc94 TH	359	{
f1f8cc94 TH	360	struct elevator_type *et = q->elevator->type;
f1f8cc94 TH	361	struct io_cq *icq;
	362
	363	/* allocate stuff */
f1f8cc94 TH	364	icq = kmem_cache_alloc_node(et->icq_cache, gfp_mask \| __GFP_ZERO,
	365	q->node);
	366	if (!icq)
	367	return NULL;
	368
	369	if (radix_tree_preload(gfp_mask) < 0) {
	370	kmem_cache_free(et->icq_cache, icq);
	371	return NULL;
	372	}
	373
	374	icq->ioc = ioc;
	375	icq->q = q;
	376	INIT_LIST_HEAD(&icq->q_node);
	377	INIT_HLIST_NODE(&icq->ioc_node);
	378
	379	/* lock both q and ioc and try to link @icq */
	380	spin_lock_irq(q->queue_lock);
	381	spin_lock(&ioc->lock);
	382
	383	if (likely(!radix_tree_insert(&ioc->icq_tree, q->id, icq))) {
	384	hlist_add_head(&icq->ioc_node, &ioc->icq_list);
	385	list_add(&icq->q_node, &q->icq_list);
	386	if (et->ops.elevator_init_icq_fn)
	387	et->ops.elevator_init_icq_fn(icq);
	388	} else {
	389	kmem_cache_free(et->icq_cache, icq);
	390	icq = ioc_lookup_icq(ioc, q);
	391	if (!icq)
	392	printk(KERN_ERR "cfq: icq link failed!\n");
	393	}
	394
	395	spin_unlock(&ioc->lock);
	396	spin_unlock_irq(q->queue_lock);
	397	radix_tree_preload_end();
	398	return icq;
	399	}
	400
13341598	401	static int __init blk_ioc_init(void)
86db1e29 JA	402	{
	403	iocontext_cachep = kmem_cache_create("blkdev_ioc",
	404	sizeof(struct io_context), 0, SLAB_PANIC, NULL);
	405	return 0;
	406	}
	407	subsys_initcall(blk_ioc_init);