[linux-block.git] / block / blk-ioc.c

// SPDX-License-Identifier: GPL-2.0
/*
 * 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/slab.h>
#include <linux/security.h>
#include <linux/sched/task.h>

#include "blk.h"
#include "blk-mq-sched.h"

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

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

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 ioc locked for blk-mq, and with both ioc
 * and queue locked for legacy.
 */
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.exit_icq)
		et->ops.exit_icq(icq);

	icq->flags |= ICQ_EXITED;
}

static void ioc_exit_icqs(struct io_context *ioc)
{
	struct io_cq *icq;

	spin_lock_irq(&ioc->lock);
	hlist_for_each_entry(icq, &ioc->icq_list, ioc_node)
		ioc_exit_icq(icq);
	spin_unlock_irq(&ioc->lock);
}

/*
 * Release an icq. Called with ioc locked for blk-mq, and with both ioc
 * and queue locked for legacy.
 */
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);

	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_access_pointer(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;
	icq->flags |= ICQ_DESTROYED;
	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);
	spin_lock_irq(&ioc->lock);

	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 {
			/* Make sure q and icq cannot be freed. */
			rcu_read_lock();

			/* Re-acquire the locks in the correct order. */
			spin_unlock(&ioc->lock);
			spin_lock(&q->queue_lock);
			spin_lock(&ioc->lock);

			/*
			 * The icq may have been destroyed when the ioc lock
			 * was released.
			 */
			if (!(icq->flags & ICQ_DESTROYED))
				ioc_destroy_icq(icq);

			spin_unlock(&q->queue_lock);
			rcu_read_unlock();
		}
	}

	spin_unlock_irq(&ioc->lock);

	kmem_cache_free(iocontext_cachep, ioc);
}

/*
 * Releasing icqs requires reverse order double locking and we may already be
 * holding a queue_lock.  Do it asynchronously from a workqueue.
 */
static bool ioc_delay_free(struct io_context *ioc)
{
	unsigned long flags;

	spin_lock_irqsave(&ioc->lock, flags);
	if (!hlist_empty(&ioc->icq_list)) {
		queue_work(system_power_efficient_wq, &ioc->release_work);
		spin_unlock_irqrestore(&ioc->lock, flags);
		return true;
	}
	spin_unlock_irqrestore(&ioc->lock, flags);
	return false;
}

/**
 * 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.
 */
void ioc_clear_queue(struct request_queue *q)
{
	LIST_HEAD(icq_list);

	spin_lock_irq(&q->queue_lock);
	list_splice_init(&q->icq_list, &icq_list);
	spin_unlock_irq(&q->queue_lock);

	rcu_read_lock();
	while (!list_empty(&icq_list)) {
		struct io_cq *icq =
			list_entry(icq_list.next, struct io_cq, q_node);

		spin_lock_irq(&icq->ioc->lock);
		if (!(icq->flags & ICQ_DESTROYED))
			ioc_destroy_icq(icq);
		spin_unlock_irq(&icq->ioc->lock);
	}
	rcu_read_unlock();
}
#else /* CONFIG_BLK_ICQ */
static inline void ioc_exit_icqs(struct io_context *ioc)
{
}
static inline bool ioc_delay_free(struct io_context *ioc)
{
	return false;
}
#endif /* CONFIG_BLK_ICQ */

/**
 * 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)
{
	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
	if (atomic_long_dec_and_test(&ioc->refcount) && !ioc_delay_free(ioc))
		kmem_cache_free(iocontext_cachep, ioc);
}
EXPORT_SYMBOL_GPL(put_io_context);

/* 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);

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

static struct io_context *alloc_io_context(gfp_t gfp_flags, int node)
{
	struct io_context *ioc;

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

	atomic_long_set(&ioc->refcount, 1);
	atomic_set(&ioc->active_ref, 1);
#ifdef CONFIG_BLK_ICQ
	spin_lock_init(&ioc->lock);
	INIT_RADIX_TREE(&ioc->icq_tree, GFP_ATOMIC);
	INIT_HLIST_HEAD(&ioc->icq_list);
	INIT_WORK(&ioc->release_work, ioc_release_fn);
#endif
	ioc->ioprio = IOPRIO_DEFAULT;

	return ioc;
}

int set_task_ioprio(struct task_struct *task, int ioprio)
{
	int err;
	const struct cred *cred = current_cred(), *tcred;

	rcu_read_lock();
	tcred = __task_cred(task);
	if (!uid_eq(tcred->uid, cred->euid) &&
	    !uid_eq(tcred->uid, cred->uid) && !capable(CAP_SYS_NICE)) {
		rcu_read_unlock();
		return -EPERM;
	}
	rcu_read_unlock();

	err = security_task_setioprio(task, ioprio);
	if (err)
		return err;

	task_lock(task);
	if (unlikely(!task->io_context)) {
		struct io_context *ioc;

		task_unlock(task);

		ioc = alloc_io_context(GFP_ATOMIC, NUMA_NO_NODE);
		if (!ioc)
			return -ENOMEM;

		task_lock(task);
		if (task->flags & PF_EXITING) {
			kmem_cache_free(iocontext_cachep, ioc);
			goto out;
		}
		if (task->io_context)
			kmem_cache_free(iocontext_cachep, ioc);
		else
			task->io_context = ioc;
	}
	task->io_context->ioprio = ioprio;
out:
	task_unlock(task);
	return 0;
}
EXPORT_SYMBOL_GPL(set_task_ioprio);

int __copy_io(unsigned long clone_flags, struct task_struct *tsk)
{
	struct io_context *ioc = current->io_context;

	/*
	 * Share io context with parent, if CLONE_IO is set
	 */
	if (clone_flags & CLONE_IO) {
		atomic_inc(&ioc->active_ref);
		tsk->io_context = ioc;
	} else if (ioprio_valid(ioc->ioprio)) {
		tsk->io_context = alloc_io_context(GFP_KERNEL, NUMA_NO_NODE);
		if (!tsk->io_context)
			return -ENOMEM;
		tsk->io_context->ioprio = ioc->ioprio;
	}

	return 0;
}

#ifdef CONFIG_BLK_ICQ
/**
 * ioc_lookup_icq - lookup io_cq from ioc
 * @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 request_queue *q)
{
	struct io_context *ioc = current->io_context;
	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
 * @q: request_queue of interest
 *
 * 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.
 */
static struct io_cq *ioc_create_icq(struct request_queue *q)
{
	struct io_context *ioc = current->io_context;
	struct elevator_type *et = q->elevator->type;
	struct io_cq *icq;

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

	if (radix_tree_maybe_preload(GFP_ATOMIC) < 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.init_icq)
			et->ops.init_icq(icq);
	} else {
		kmem_cache_free(et->icq_cache, icq);
		icq = ioc_lookup_icq(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;
}

struct io_cq *ioc_find_get_icq(struct request_queue *q)
{
	struct io_context *ioc = current->io_context;
	struct io_cq *icq = NULL;

	if (unlikely(!ioc)) {
		ioc = alloc_io_context(GFP_ATOMIC, q->node);
		if (!ioc)
			return NULL;

		task_lock(current);
		if (current->io_context) {
			kmem_cache_free(iocontext_cachep, ioc);
			ioc = current->io_context;
		} else {
			current->io_context = ioc;
		}

		get_io_context(ioc);
		task_unlock(current);
	} else {
		get_io_context(ioc);

		spin_lock_irq(&q->queue_lock);
		icq = ioc_lookup_icq(q);
		spin_unlock_irq(&q->queue_lock);
	}

	if (!icq) {
		icq = ioc_create_icq(q);
		if (!icq) {
			put_io_context(ioc);
			return NULL;
		}
	}
	return icq;
}
EXPORT_SYMBOL_GPL(ioc_find_get_icq);
#endif /* CONFIG_BLK_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
b2441318	1	// SPDX-License-Identifier: GPL-2.0
86db1e29 JA	2	/*
	3	* Functions related to io context handling
	4	*/
	5	#include <linux/kernel.h>
	6	#include <linux/module.h>
	7	#include <linux/init.h>
	8	#include <linux/bio.h>
	9	#include <linux/blkdev.h>
5a0e3ad6	10	#include <linux/slab.h>
a411cd3c	11	#include <linux/security.h>
f719ff9b	12	#include <linux/sched/task.h>
86db1e29 JA	13
86db1e29 JA	14	#include "blk.h"
2aa7745b	15	#include "blk-mq-sched.h"
86db1e29 JA	16
	17	/*
	18	* For io context allocations
	19	*/
	20	static struct kmem_cache *iocontext_cachep;
	21
5ef16305	22	#ifdef CONFIG_BLK_ICQ
6e736be7 TH	23	/**
	24	* get_io_context - increment reference count to io_context
	25	* @ioc: io_context to get
	26	*
	27	* Increment reference count to @ioc.
	28	*/
87dd1d63	29	static void get_io_context(struct io_context *ioc)
6e736be7 TH	30	{
	31	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
	32	atomic_long_inc(&ioc->refcount);
	33	}
6e736be7	34
7e5a8794 TH	35	static void icq_free_icq_rcu(struct rcu_head *head)
	36	{
	37	struct io_cq *icq = container_of(head, struct io_cq, __rcu_head);
	38
	39	kmem_cache_free(icq->__rcu_icq_cache, icq);
	40	}
	41
3d492c2e	42	/*
7b36a718 JA	43	* Exit an icq. Called with ioc locked for blk-mq, and with both ioc
7b36a718 JA	44	* and queue locked for legacy.
3d492c2e	45	*/
7e5a8794	46	static void ioc_exit_icq(struct io_cq *icq)
621032ad TH	47	{
	48	struct elevator_type *et = icq->q->elevator->type;
	49
	50	if (icq->flags & ICQ_EXITED)
	51	return;
	52
f9cd4bfe JA	53	if (et->ops.exit_icq)
f9cd4bfe JA	54	et->ops.exit_icq(icq);
621032ad TH	55
	56	icq->flags \|= ICQ_EXITED;
	57	}
	58
4be8a2ea CH	59	static void ioc_exit_icqs(struct io_context *ioc)
	60	{
	61	struct io_cq *icq;
	62
	63	spin_lock_irq(&ioc->lock);
	64	hlist_for_each_entry(icq, &ioc->icq_list, ioc_node)
	65	ioc_exit_icq(icq);
	66	spin_unlock_irq(&ioc->lock);
	67	}
	68
7b36a718 JA	69	/*
	70	* Release an icq. Called with ioc locked for blk-mq, and with both ioc
	71	* and queue locked for legacy.
	72	*/
621032ad	73	static void ioc_destroy_icq(struct io_cq *icq)
7e5a8794 TH	74	{
	75	struct io_context *ioc = icq->ioc;
	76	struct request_queue *q = icq->q;
	77	struct elevator_type *et = q->elevator->type;
	78
	79	lockdep_assert_held(&ioc->lock);
7e5a8794 TH	80
	81	radix_tree_delete(&ioc->icq_tree, icq->q->id);
	82	hlist_del_init(&icq->ioc_node);
	83	list_del_init(&icq->q_node);
	84
	85	/*
	86	* Both setting lookup hint to and clearing it from @icq are done
	87	* under queue_lock. If it's not pointing to @icq now, it never
	88	* will. Hint assignment itself can race safely.
	89	*/
ec6c676a	90	if (rcu_access_pointer(ioc->icq_hint) == icq)
7e5a8794 TH	91	rcu_assign_pointer(ioc->icq_hint, NULL);
7e5a8794 TH	92
621032ad	93	ioc_exit_icq(icq);
7e5a8794 TH	94
	95	/*
	96	* @icq->q might have gone away by the time RCU callback runs
	97	* making it impossible to determine icq_cache. Record it in @icq.
	98	*/
	99	icq->__rcu_icq_cache = et->icq_cache;
30a2da7b	100	icq->flags \|= ICQ_DESTROYED;
7e5a8794 TH	101	call_rcu(&icq->__rcu_head, icq_free_icq_rcu);
	102	}
	103
b2efa052 TH	104	/*
b2efa052 TH	105	* Slow path for ioc release in put_io_context(). Performs double-lock
c5869807	106	* dancing to unlink all icq's and then frees ioc.
b2efa052 TH	107	*/
b2efa052 TH	108	static void ioc_release_fn(struct work_struct *work)
86db1e29	109	{
b2efa052 TH	110	struct io_context *ioc = container_of(work, struct io_context,
b2efa052 TH	111	release_work);
a43f085f	112	spin_lock_irq(&ioc->lock);
b2efa052	113
c5869807 TH	114	while (!hlist_empty(&ioc->icq_list)) {
	115	struct io_cq *icq = hlist_entry(ioc->icq_list.first,
	116	struct io_cq, ioc_node);
2274b029 TH	117	struct request_queue *q = icq->q;
2274b029 TH	118
0d945c1f	119	if (spin_trylock(&q->queue_lock)) {
621032ad	120	ioc_destroy_icq(icq);
0d945c1f	121	spin_unlock(&q->queue_lock);
2274b029	122	} else {
ab96bbab JO	123	/* Make sure q and icq cannot be freed. */
	124	rcu_read_lock();
	125
	126	/* Re-acquire the locks in the correct order. */
	127	spin_unlock(&ioc->lock);
	128	spin_lock(&q->queue_lock);
	129	spin_lock(&ioc->lock);
	130
	131	/*
	132	* The icq may have been destroyed when the ioc lock
	133	* was released.
	134	*/
	135	if (!(icq->flags & ICQ_DESTROYED))
	136	ioc_destroy_icq(icq);
	137
	138	spin_unlock(&q->queue_lock);
	139	rcu_read_unlock();
b2efa052	140	}
b2efa052	141	}
ffc4e759	142
a43f085f	143	spin_unlock_irq(&ioc->lock);
b2efa052 TH	144
b2efa052 TH	145	kmem_cache_free(iocontext_cachep, ioc);
86db1e29 JA	146	}
86db1e29 JA	147
edf70ff5 CH	148	/*
	149	* Releasing icqs requires reverse order double locking and we may already be
	150	* holding a queue_lock. Do it asynchronously from a workqueue.
	151	*/
	152	static bool ioc_delay_free(struct io_context *ioc)
	153	{
	154	unsigned long flags;
	155
	156	spin_lock_irqsave(&ioc->lock, flags);
	157	if (!hlist_empty(&ioc->icq_list)) {
	158	queue_work(system_power_efficient_wq, &ioc->release_work);
	159	spin_unlock_irqrestore(&ioc->lock, flags);
	160	return true;
	161	}
	162	spin_unlock_irqrestore(&ioc->lock, flags);
	163	return false;
	164	}
	165
5ef16305 CH	166	/**
	167	* ioc_clear_queue - break any ioc association with the specified queue
	168	* @q: request_queue being cleared
	169	*
	170	* Walk @q->icq_list and exit all io_cq's.
	171	*/
	172	void ioc_clear_queue(struct request_queue *q)
	173	{
	174	LIST_HEAD(icq_list);
	175
	176	spin_lock_irq(&q->queue_lock);
	177	list_splice_init(&q->icq_list, &icq_list);
	178	spin_unlock_irq(&q->queue_lock);
	179
	180	rcu_read_lock();
	181	while (!list_empty(&icq_list)) {
	182	struct io_cq *icq =
	183	list_entry(icq_list.next, struct io_cq, q_node);
	184
	185	spin_lock_irq(&icq->ioc->lock);
	186	if (!(icq->flags & ICQ_DESTROYED))
	187	ioc_destroy_icq(icq);
	188	spin_unlock_irq(&icq->ioc->lock);
	189	}
	190	rcu_read_unlock();
	191	}
	192	#else /* CONFIG_BLK_ICQ */
	193	static inline void ioc_exit_icqs(struct io_context *ioc)
	194	{
	195	}
	196	static inline bool ioc_delay_free(struct io_context *ioc)
	197	{
	198	return false;
	199	}
	200	#endif /* CONFIG_BLK_ICQ */
	201
42ec57a8 TH	202	/**
	203	* put_io_context - put a reference of io_context
	204	* @ioc: io_context to put
	205	*
	206	* Decrement reference count of @ioc and release it if the count reaches
11a3122f	207	* zero.
86db1e29	208	*/
11a3122f	209	void put_io_context(struct io_context *ioc)
86db1e29	210	{
42ec57a8	211	BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
edf70ff5	212	if (atomic_long_dec_and_test(&ioc->refcount) && !ioc_delay_free(ioc))
ff8c1474	213	kmem_cache_free(iocontext_cachep, ioc);
86db1e29	214	}
222ee581	215	EXPORT_SYMBOL_GPL(put_io_context);
86db1e29	216
f6e8d01b TH	217	/* Called by the exiting task */
	218	void exit_io_context(struct task_struct *task)
	219	{
	220	struct io_context *ioc;
	221
	222	task_lock(task);
	223	ioc = task->io_context;
	224	task->io_context = NULL;
	225	task_unlock(task);
	226
4be8a2ea CH	227	if (atomic_dec_and_test(&ioc->active_ref)) {
	228	ioc_exit_icqs(ioc);
	229	put_io_context(ioc);
	230	}
f6e8d01b TH	231	}
f6e8d01b TH	232
a0f14d8b	233	static struct io_context *alloc_io_context(gfp_t gfp_flags, int node)
86db1e29	234	{
df415656	235	struct io_context *ioc;
86db1e29	236
42ec57a8 TH	237	ioc = kmem_cache_alloc_node(iocontext_cachep, gfp_flags \| __GFP_ZERO,
	238	node);
	239	if (unlikely(!ioc))
a0f14d8b	240	return NULL;
42ec57a8	241
42ec57a8	242	atomic_long_set(&ioc->refcount, 1);
f6e8d01b	243	atomic_set(&ioc->active_ref, 1);
5ef16305	244	#ifdef CONFIG_BLK_ICQ
42ec57a8	245	spin_lock_init(&ioc->lock);
c137969b	246	INIT_RADIX_TREE(&ioc->icq_tree, GFP_ATOMIC);
c5869807	247	INIT_HLIST_HEAD(&ioc->icq_list);
b2efa052	248	INIT_WORK(&ioc->release_work, ioc_release_fn);
5ef16305	249	#endif
e589f464 JK	250	ioc->ioprio = IOPRIO_DEFAULT;
e589f464 JK	251
a0f14d8b CH	252	return ioc;
	253	}
	254
a411cd3c CH	255	int set_task_ioprio(struct task_struct *task, int ioprio)
	256	{
	257	int err;
a411cd3c CH	258	const struct cred cred = current_cred(), tcred;
	259
	260	rcu_read_lock();
	261	tcred = __task_cred(task);
	262	if (!uid_eq(tcred->uid, cred->euid) &&
	263	!uid_eq(tcred->uid, cred->uid) && !capable(CAP_SYS_NICE)) {
	264	rcu_read_unlock();
	265	return -EPERM;
	266	}
	267	rcu_read_unlock();
	268
	269	err = security_task_setioprio(task, ioprio);
	270	if (err)
	271	return err;
	272
8472161b CH	273	task_lock(task);
	274	if (unlikely(!task->io_context)) {
	275	struct io_context *ioc;
a411cd3c	276
8472161b	277	task_unlock(task);
5fc11eeb CH	278
	279	ioc = alloc_io_context(GFP_ATOMIC, NUMA_NO_NODE);
	280	if (!ioc)
	281	return -ENOMEM;
	282
	283	task_lock(task);
a957b612	284	if (task->flags & PF_EXITING) {
a957b612 JA	285	kmem_cache_free(iocontext_cachep, ioc);
	286	goto out;
	287	}
669a0646	288	if (task->io_context)
5fc11eeb	289	kmem_cache_free(iocontext_cachep, ioc);
669a0646	290	else
5fc11eeb	291	task->io_context = ioc;
8472161b CH	292	}
8472161b CH	293	task->io_context->ioprio = ioprio;
a957b612	294	out:
8472161b	295	task_unlock(task);
15583a56	296	return 0;
a411cd3c CH	297	}
	298	EXPORT_SYMBOL_GPL(set_task_ioprio);
	299
88c9a2ce CH	300	int __copy_io(unsigned long clone_flags, struct task_struct *tsk)
	301	{
	302	struct io_context *ioc = current->io_context;
88c9a2ce CH	303
	304	/*
	305	* Share io context with parent, if CLONE_IO is set
	306	*/
	307	if (clone_flags & CLONE_IO) {
50569c24	308	atomic_inc(&ioc->active_ref);
88c9a2ce CH	309	tsk->io_context = ioc;
88c9a2ce CH	310	} else if (ioprio_valid(ioc->ioprio)) {
8ffc1368 CH	311	tsk->io_context = alloc_io_context(GFP_KERNEL, NUMA_NO_NODE);
8ffc1368 CH	312	if (!tsk->io_context)
88c9a2ce	313	return -ENOMEM;
8ffc1368	314	tsk->io_context->ioprio = ioc->ioprio;
88c9a2ce CH	315	}
	316
	317	return 0;
	318	}
	319
5ef16305	320	#ifdef CONFIG_BLK_ICQ
47fdd4ca TH	321	/**
47fdd4ca TH	322	* ioc_lookup_icq - lookup io_cq from ioc
47fdd4ca TH	323	* @q: the associated request_queue
	324	*
	325	* Look up io_cq associated with @ioc - @q pair from @ioc. Must be called
	326	* with @q->queue_lock held.
	327	*/
eca5892a	328	struct io_cq ioc_lookup_icq(struct request_queue q)
47fdd4ca	329	{
eca5892a	330	struct io_context *ioc = current->io_context;
47fdd4ca TH	331	struct io_cq *icq;
47fdd4ca TH	332
0d945c1f	333	lockdep_assert_held(&q->queue_lock);
47fdd4ca TH	334
	335	/*
	336	* icq's are indexed from @ioc using radix tree and hint pointer,
	337	* both of which are protected with RCU. All removals are done
	338	* holding both q and ioc locks, and we're holding q lock - if we
	339	* find a icq which points to us, it's guaranteed to be valid.
	340	*/
	341	rcu_read_lock();
	342	icq = rcu_dereference(ioc->icq_hint);
	343	if (icq && icq->q == q)
	344	goto out;
	345
	346	icq = radix_tree_lookup(&ioc->icq_tree, q->id);
	347	if (icq && icq->q == q)
	348	rcu_assign_pointer(ioc->icq_hint, icq); /* allowed to race */
	349	else
	350	icq = NULL;
	351	out:
	352	rcu_read_unlock();
	353	return icq;
	354	}
	355	EXPORT_SYMBOL(ioc_lookup_icq);
	356
f1f8cc94 TH	357	/**
	358	* ioc_create_icq - create and link io_cq
	359	* @q: request_queue of interest
f1f8cc94	360	*
24acfc34 TH	361	* Make sure io_cq linking @ioc and @q exists. If icq doesn't exist, they
24acfc34 TH	362	* will be created using @gfp_mask.
f1f8cc94 TH	363	*
	364	* The caller is responsible for ensuring @ioc won't go away and @q is
	365	* alive and will stay alive until this function returns.
	366	*/
18b74c4d	367	static struct io_cq ioc_create_icq(struct request_queue q)
f1f8cc94	368	{
18b74c4d	369	struct io_context *ioc = current->io_context;
f1f8cc94	370	struct elevator_type *et = q->elevator->type;
f1f8cc94 TH	371	struct io_cq *icq;
	372
	373	/* allocate stuff */
18b74c4d	374	icq = kmem_cache_alloc_node(et->icq_cache, GFP_ATOMIC \| __GFP_ZERO,
f1f8cc94 TH	375	q->node);
	376	if (!icq)
	377	return NULL;
	378
18b74c4d	379	if (radix_tree_maybe_preload(GFP_ATOMIC) < 0) {
f1f8cc94 TH	380	kmem_cache_free(et->icq_cache, icq);
	381	return NULL;
	382	}
	383
	384	icq->ioc = ioc;
	385	icq->q = q;
	386	INIT_LIST_HEAD(&icq->q_node);
	387	INIT_HLIST_NODE(&icq->ioc_node);
	388
	389	/* lock both q and ioc and try to link @icq */
0d945c1f	390	spin_lock_irq(&q->queue_lock);
f1f8cc94 TH	391	spin_lock(&ioc->lock);
	392
	393	if (likely(!radix_tree_insert(&ioc->icq_tree, q->id, icq))) {
	394	hlist_add_head(&icq->ioc_node, &ioc->icq_list);
	395	list_add(&icq->q_node, &q->icq_list);
f9cd4bfe JA	396	if (et->ops.init_icq)
f9cd4bfe JA	397	et->ops.init_icq(icq);
f1f8cc94 TH	398	} else {
f1f8cc94 TH	399	kmem_cache_free(et->icq_cache, icq);
eca5892a	400	icq = ioc_lookup_icq(q);
f1f8cc94 TH	401	if (!icq)
	402	printk(KERN_ERR "cfq: icq link failed!\n");
	403	}
	404
	405	spin_unlock(&ioc->lock);
0d945c1f	406	spin_unlock_irq(&q->queue_lock);
f1f8cc94 TH	407	radix_tree_preload_end();
	408	return icq;
	409	}
	410
87dd1d63 CH	411	struct io_cq ioc_find_get_icq(struct request_queue q)
87dd1d63 CH	412	{
d538ea4c CH	413	struct io_context *ioc = current->io_context;
d538ea4c CH	414	struct io_cq *icq = NULL;
87dd1d63	415
d538ea4c	416	if (unlikely(!ioc)) {
90b627f5	417	ioc = alloc_io_context(GFP_ATOMIC, q->node);
d538ea4c CH	418	if (!ioc)
d538ea4c CH	419	return NULL;
90b627f5 CH	420
	421	task_lock(current);
	422	if (current->io_context) {
	423	kmem_cache_free(iocontext_cachep, ioc);
	424	ioc = current->io_context;
	425	} else {
	426	current->io_context = ioc;
	427	}
	428
	429	get_io_context(ioc);
	430	task_unlock(current);
d538ea4c CH	431	} else {
d538ea4c CH	432	get_io_context(ioc);
87dd1d63	433
d538ea4c	434	spin_lock_irq(&q->queue_lock);
eca5892a	435	icq = ioc_lookup_icq(q);
d538ea4c CH	436	spin_unlock_irq(&q->queue_lock);
d538ea4c CH	437	}
87dd1d63 CH	438
87dd1d63 CH	439	if (!icq) {
18b74c4d	440	icq = ioc_create_icq(q);
d538ea4c CH	441	if (!icq) {
d538ea4c CH	442	put_io_context(ioc);
87dd1d63	443	return NULL;
d538ea4c	444	}
87dd1d63	445	}
87dd1d63 CH	446	return icq;
	447	}
	448	EXPORT_SYMBOL_GPL(ioc_find_get_icq);
5ef16305	449	#endif /* CONFIG_BLK_ICQ */
87dd1d63	450
13341598	451	static int __init blk_ioc_init(void)
86db1e29 JA	452	{
	453	iocontext_cachep = kmem_cache_create("blkdev_ioc",
	454	sizeof(struct io_context), 0, SLAB_PANIC, NULL);
	455	return 0;
	456	}
	457	subsys_initcall(blk_ioc_init);