[linux-block.git] / io_uring / io_uring.h

#ifndef IOU_CORE_H
#define IOU_CORE_H

#include <linux/errno.h>
#include <linux/lockdep.h>
#include <linux/io_uring_types.h>
#include <uapi/linux/eventpoll.h>
#include "io-wq.h"
#include "slist.h"
#include "filetable.h"

#ifndef CREATE_TRACE_POINTS
#include <trace/events/io_uring.h>
#endif

enum {
	IOU_OK			= 0,
	IOU_ISSUE_SKIP_COMPLETE	= -EIOCBQUEUED,

	/*
	 * Intended only when both IO_URING_F_MULTISHOT is passed
	 * to indicate to the poll runner that multishot should be
	 * removed and the result is set on req->cqe.res.
	 */
	IOU_STOP_MULTISHOT	= -ECANCELED,
};

struct io_uring_cqe *__io_get_cqe(struct io_ring_ctx *ctx, bool overflow);
bool io_req_cqe_overflow(struct io_kiocb *req);
int io_run_task_work_sig(struct io_ring_ctx *ctx);
int __io_run_local_work(struct io_ring_ctx *ctx, bool *locked);
int io_run_local_work(struct io_ring_ctx *ctx);
void io_req_complete_failed(struct io_kiocb *req, s32 res);
void __io_req_complete(struct io_kiocb *req, unsigned issue_flags);
void io_req_complete_post(struct io_kiocb *req);
bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
		     bool allow_overflow);
bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
		     bool allow_overflow);
void __io_commit_cqring_flush(struct io_ring_ctx *ctx);

struct page **io_pin_pages(unsigned long ubuf, unsigned long len, int *npages);

struct file *io_file_get_normal(struct io_kiocb *req, int fd);
struct file *io_file_get_fixed(struct io_kiocb *req, int fd,
			       unsigned issue_flags);

static inline bool io_req_ffs_set(struct io_kiocb *req)
{
	return req->flags & REQ_F_FIXED_FILE;
}

void __io_req_task_work_add(struct io_kiocb *req, bool allow_local);
bool io_is_uring_fops(struct file *file);
bool io_alloc_async_data(struct io_kiocb *req);
void io_req_task_queue(struct io_kiocb *req);
void io_queue_iowq(struct io_kiocb *req, bool *dont_use);
void io_req_task_complete(struct io_kiocb *req, bool *locked);
void io_req_task_queue_fail(struct io_kiocb *req, int ret);
void io_req_task_submit(struct io_kiocb *req, bool *locked);
void tctx_task_work(struct callback_head *cb);
__cold void io_uring_cancel_generic(bool cancel_all, struct io_sq_data *sqd);
int io_uring_alloc_task_context(struct task_struct *task,
				struct io_ring_ctx *ctx);

int io_poll_issue(struct io_kiocb *req, bool *locked);
int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr);
int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin);
void io_free_batch_list(struct io_ring_ctx *ctx, struct io_wq_work_node *node);
int io_req_prep_async(struct io_kiocb *req);

struct io_wq_work *io_wq_free_work(struct io_wq_work *work);
void io_wq_submit_work(struct io_wq_work *work);

void io_free_req(struct io_kiocb *req);
void io_queue_next(struct io_kiocb *req);
void __io_put_task(struct task_struct *task, int nr);
void io_task_refs_refill(struct io_uring_task *tctx);
bool __io_alloc_req_refill(struct io_ring_ctx *ctx);

bool io_match_task_safe(struct io_kiocb *head, struct task_struct *task,
			bool cancel_all);

static inline void io_req_task_work_add(struct io_kiocb *req)
{
	__io_req_task_work_add(req, true);
}

#define io_for_each_link(pos, head) \
	for (pos = (head); pos; pos = pos->link)

static inline void io_cq_lock(struct io_ring_ctx *ctx)
	__acquires(ctx->completion_lock)
{
	spin_lock(&ctx->completion_lock);
}

void io_cq_unlock_post(struct io_ring_ctx *ctx);

static inline struct io_uring_cqe *io_get_cqe_overflow(struct io_ring_ctx *ctx,
						       bool overflow)
{
	if (likely(ctx->cqe_cached < ctx->cqe_sentinel)) {
		struct io_uring_cqe *cqe = ctx->cqe_cached;

		ctx->cached_cq_tail++;
		ctx->cqe_cached++;
		if (ctx->flags & IORING_SETUP_CQE32)
			ctx->cqe_cached++;
		return cqe;
	}

	return __io_get_cqe(ctx, overflow);
}

static inline struct io_uring_cqe *io_get_cqe(struct io_ring_ctx *ctx)
{
	return io_get_cqe_overflow(ctx, false);
}

static inline bool __io_fill_cqe_req(struct io_ring_ctx *ctx,
				     struct io_kiocb *req)
{
	struct io_uring_cqe *cqe;

	/*
	 * If we can't get a cq entry, userspace overflowed the
	 * submission (by quite a lot). Increment the overflow count in
	 * the ring.
	 */
	cqe = io_get_cqe(ctx);
	if (unlikely(!cqe))
		return io_req_cqe_overflow(req);

	trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
				req->cqe.res, req->cqe.flags,
				(req->flags & REQ_F_CQE32_INIT) ? req->extra1 : 0,
				(req->flags & REQ_F_CQE32_INIT) ? req->extra2 : 0);

	memcpy(cqe, &req->cqe, sizeof(*cqe));

	if (ctx->flags & IORING_SETUP_CQE32) {
		u64 extra1 = 0, extra2 = 0;

		if (req->flags & REQ_F_CQE32_INIT) {
			extra1 = req->extra1;
			extra2 = req->extra2;
		}

		WRITE_ONCE(cqe->big_cqe[0], extra1);
		WRITE_ONCE(cqe->big_cqe[1], extra2);
	}
	return true;
}

static inline void req_set_fail(struct io_kiocb *req)
{
	req->flags |= REQ_F_FAIL;
	if (req->flags & REQ_F_CQE_SKIP) {
		req->flags &= ~REQ_F_CQE_SKIP;
		req->flags |= REQ_F_SKIP_LINK_CQES;
	}
}

static inline void io_req_set_res(struct io_kiocb *req, s32 res, u32 cflags)
{
	req->cqe.res = res;
	req->cqe.flags = cflags;
}

static inline bool req_has_async_data(struct io_kiocb *req)
{
	return req->flags & REQ_F_ASYNC_DATA;
}

static inline void io_put_file(struct file *file)
{
	if (file)
		fput(file);
}

static inline void io_ring_submit_unlock(struct io_ring_ctx *ctx,
					 unsigned issue_flags)
{
	lockdep_assert_held(&ctx->uring_lock);
	if (issue_flags & IO_URING_F_UNLOCKED)
		mutex_unlock(&ctx->uring_lock);
}

static inline void io_ring_submit_lock(struct io_ring_ctx *ctx,
				       unsigned issue_flags)
{
	/*
	 * "Normal" inline submissions always hold the uring_lock, since we
	 * grab it from the system call. Same is true for the SQPOLL offload.
	 * The only exception is when we've detached the request and issue it
	 * from an async worker thread, grab the lock for that case.
	 */
	if (issue_flags & IO_URING_F_UNLOCKED)
		mutex_lock(&ctx->uring_lock);
	lockdep_assert_held(&ctx->uring_lock);
}

static inline void io_commit_cqring(struct io_ring_ctx *ctx)
{
	/* order cqe stores with ring update */
	smp_store_release(&ctx->rings->cq.tail, ctx->cached_cq_tail);
}

/* requires smb_mb() prior, see wq_has_sleeper() */
static inline void __io_cqring_wake(struct io_ring_ctx *ctx)
{
	/*
	 * Trigger waitqueue handler on all waiters on our waitqueue. This
	 * won't necessarily wake up all the tasks, io_should_wake() will make
	 * that decision.
	 *
	 * Pass in EPOLLIN|EPOLL_URING_WAKE as the poll wakeup key. The latter
	 * set in the mask so that if we recurse back into our own poll
	 * waitqueue handlers, we know we have a dependency between eventfd or
	 * epoll and should terminate multishot poll at that point.
	 */
	if (waitqueue_active(&ctx->cq_wait))
		__wake_up(&ctx->cq_wait, TASK_NORMAL, 0,
				poll_to_key(EPOLL_URING_WAKE | EPOLLIN));
}

static inline void io_cqring_wake(struct io_ring_ctx *ctx)
{
	smp_mb();
	__io_cqring_wake(ctx);
}

static inline bool io_sqring_full(struct io_ring_ctx *ctx)
{
	struct io_rings *r = ctx->rings;

	return READ_ONCE(r->sq.tail) - ctx->cached_sq_head == ctx->sq_entries;
}

static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
{
	struct io_rings *rings = ctx->rings;

	/* make sure SQ entry isn't read before tail */
	return smp_load_acquire(&rings->sq.tail) - ctx->cached_sq_head;
}

static inline int io_run_task_work(void)
{
	if (task_work_pending(current)) {
		if (test_thread_flag(TIF_NOTIFY_SIGNAL))
			clear_notify_signal();
		__set_current_state(TASK_RUNNING);
		task_work_run();
		return 1;
	}

	return 0;
}

static inline bool io_task_work_pending(struct io_ring_ctx *ctx)
{
	return test_thread_flag(TIF_NOTIFY_SIGNAL) ||
		!wq_list_empty(&ctx->work_llist);
}

static inline int io_run_task_work_ctx(struct io_ring_ctx *ctx)
{
	int ret = 0;
	int ret2;

	if (ctx->flags & IORING_SETUP_DEFER_TASKRUN)
		ret = io_run_local_work(ctx);

	/* want to run this after in case more is added */
	ret2 = io_run_task_work();

	/* Try propagate error in favour of if tasks were run,
	 * but still make sure to run them if requested
	 */
	if (ret >= 0)
		ret += ret2;

	return ret;
}

static inline int io_run_local_work_locked(struct io_ring_ctx *ctx)
{
	bool locked;
	int ret;

	if (llist_empty(&ctx->work_llist))
		return 0;

	locked = true;
	ret = __io_run_local_work(ctx, &locked);
	/* shouldn't happen! */
	if (WARN_ON_ONCE(!locked))
		mutex_lock(&ctx->uring_lock);
	return ret;
}

static inline void io_tw_lock(struct io_ring_ctx *ctx, bool *locked)
{
	if (!*locked) {
		mutex_lock(&ctx->uring_lock);
		*locked = true;
	}
}

/*
 * Don't complete immediately but use deferred completion infrastructure.
 * Protected by ->uring_lock and can only be used either with
 * IO_URING_F_COMPLETE_DEFER or inside a tw handler holding the mutex.
 */
static inline void io_req_complete_defer(struct io_kiocb *req)
	__must_hold(&req->ctx->uring_lock)
{
	struct io_submit_state *state = &req->ctx->submit_state;

	lockdep_assert_held(&req->ctx->uring_lock);

	wq_list_add_tail(&req->comp_list, &state->compl_reqs);
}

static inline void io_commit_cqring_flush(struct io_ring_ctx *ctx)
{
	if (unlikely(ctx->off_timeout_used || ctx->drain_active || ctx->has_evfd))
		__io_commit_cqring_flush(ctx);
}

/* must to be called somewhat shortly after putting a request */
static inline void io_put_task(struct task_struct *task, int nr)
{
	if (likely(task == current))
		task->io_uring->cached_refs += nr;
	else
		__io_put_task(task, nr);
}

static inline void io_get_task_refs(int nr)
{
	struct io_uring_task *tctx = current->io_uring;

	tctx->cached_refs -= nr;
	if (unlikely(tctx->cached_refs < 0))
		io_task_refs_refill(tctx);
}

static inline bool io_req_cache_empty(struct io_ring_ctx *ctx)
{
	return !ctx->submit_state.free_list.next;
}

static inline bool io_alloc_req_refill(struct io_ring_ctx *ctx)
{
	if (unlikely(io_req_cache_empty(ctx)))
		return __io_alloc_req_refill(ctx);
	return true;
}

static inline struct io_kiocb *io_alloc_req(struct io_ring_ctx *ctx)
{
	struct io_wq_work_node *node;

	node = wq_stack_extract(&ctx->submit_state.free_list);
	return container_of(node, struct io_kiocb, comp_list);
}

static inline bool io_allowed_run_tw(struct io_ring_ctx *ctx)
{
	return likely(!(ctx->flags & IORING_SETUP_DEFER_TASKRUN) ||
		      ctx->submitter_task == current);
}

static inline void io_req_queue_tw_complete(struct io_kiocb *req, s32 res)
{
	io_req_set_res(req, res, 0);
	req->io_task_work.func = io_req_task_complete;
	io_req_task_work_add(req);
}

#endif
Commit	Line	Data
de23077e JA	1	#ifndef IOU_CORE_H
	2	#define IOU_CORE_H
	3
	4	#include <linux/errno.h>
cd40cae2	5	#include <linux/lockdep.h>
ab1c84d8	6	#include <linux/io_uring_types.h>
44648532	7	#include <uapi/linux/eventpoll.h>
ab1c84d8	8	#include "io-wq.h"
a6b21fbb	9	#include "slist.h"
ab1c84d8	10	#include "filetable.h"
de23077e	11
f3b44f92 JA	12	#ifndef CREATE_TRACE_POINTS
	13	#include <trace/events/io_uring.h>
	14	#endif
	15
97b388d7 JA	16	enum {
	17	IOU_OK = 0,
	18	IOU_ISSUE_SKIP_COMPLETE = -EIOCBQUEUED,
114eccdf DY	19
114eccdf DY	20	/*
91482864 PB	21	* Intended only when both IO_URING_F_MULTISHOT is passed
91482864 PB	22	* to indicate to the poll runner that multishot should be
114eccdf DY	23	* removed and the result is set on req->cqe.res.
	24	*/
	25	IOU_STOP_MULTISHOT = -ECANCELED,
97b388d7 JA	26	};
97b388d7 JA	27
aa1df3a3	28	struct io_uring_cqe __io_get_cqe(struct io_ring_ctx ctx, bool overflow);
68494a65	29	bool io_req_cqe_overflow(struct io_kiocb *req);
c0e0d6ba	30	int io_run_task_work_sig(struct io_ring_ctx *ctx);
b3026767	31	int __io_run_local_work(struct io_ring_ctx ctx, bool locked);
c0e0d6ba	32	int io_run_local_work(struct io_ring_ctx *ctx);
9046c641 PB	33	void io_req_complete_failed(struct io_kiocb *req, s32 res);
	34	void __io_req_complete(struct io_kiocb *req, unsigned issue_flags);
	35	void io_req_complete_post(struct io_kiocb *req);
52120f0f DY	36	bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
52120f0f DY	37	bool allow_overflow);
eb42cebb PB	38	bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
eb42cebb PB	39	bool allow_overflow);
9046c641 PB	40	void __io_commit_cqring_flush(struct io_ring_ctx *ctx);
	41
	42	struct page *io_pin_pages(unsigned long ubuf, unsigned long len, int npages);
	43
	44	struct file io_file_get_normal(struct io_kiocb req, int fd);
	45	struct file io_file_get_fixed(struct io_kiocb req, int fd,
	46	unsigned issue_flags);
	47
f6b543fd JA	48	static inline bool io_req_ffs_set(struct io_kiocb *req)
	49	{
	50	return req->flags & REQ_F_FIXED_FILE;
	51	}
	52
e52d2e58	53	void __io_req_task_work_add(struct io_kiocb *req, bool allow_local);
9046c641 PB	54	bool io_is_uring_fops(struct file *file);
9046c641 PB	55	bool io_alloc_async_data(struct io_kiocb *req);
9046c641 PB	56	void io_req_task_queue(struct io_kiocb *req);
	57	void io_queue_iowq(struct io_kiocb req, bool dont_use);
	58	void io_req_task_complete(struct io_kiocb req, bool locked);
	59	void io_req_task_queue_fail(struct io_kiocb *req, int ret);
	60	void io_req_task_submit(struct io_kiocb req, bool locked);
	61	void tctx_task_work(struct callback_head *cb);
	62	__cold void io_uring_cancel_generic(bool cancel_all, struct io_sq_data *sqd);
	63	int io_uring_alloc_task_context(struct task_struct *task,
	64	struct io_ring_ctx *ctx);
	65
	66	int io_poll_issue(struct io_kiocb req, bool locked);
	67	int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr);
	68	int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin);
	69	void io_free_batch_list(struct io_ring_ctx ctx, struct io_wq_work_node node);
	70	int io_req_prep_async(struct io_kiocb *req);
	71
	72	struct io_wq_work io_wq_free_work(struct io_wq_work work);
	73	void io_wq_submit_work(struct io_wq_work *work);
	74
	75	void io_free_req(struct io_kiocb *req);
	76	void io_queue_next(struct io_kiocb *req);
e70cb608	77	void __io_put_task(struct task_struct *task, int nr);
63809137	78	void io_task_refs_refill(struct io_uring_task *tctx);
bd1a3783	79	bool __io_alloc_req_refill(struct io_ring_ctx *ctx);
9046c641 PB	80
	81	bool io_match_task_safe(struct io_kiocb head, struct task_struct task,
	82	bool cancel_all);
	83
e52d2e58 PB	84	static inline void io_req_task_work_add(struct io_kiocb *req)
	85	{
	86	__io_req_task_work_add(req, true);
	87	}
	88
9046c641 PB	89	#define io_for_each_link(pos, head) \
9046c641 PB	90	for (pos = (head); pos; pos = pos->link)
f3b44f92	91
25399321 PB	92	static inline void io_cq_lock(struct io_ring_ctx *ctx)
	93	__acquires(ctx->completion_lock)
	94	{
	95	spin_lock(&ctx->completion_lock);
	96	}
	97
	98	void io_cq_unlock_post(struct io_ring_ctx *ctx);
	99
aa1df3a3 PB	100	static inline struct io_uring_cqe io_get_cqe_overflow(struct io_ring_ctx ctx,
aa1df3a3 PB	101	bool overflow)
f3b44f92 JA	102	{
	103	if (likely(ctx->cqe_cached < ctx->cqe_sentinel)) {
	104	struct io_uring_cqe *cqe = ctx->cqe_cached;
	105
f3b44f92 JA	106	ctx->cached_cq_tail++;
f3b44f92 JA	107	ctx->cqe_cached++;
b3659a65 PB	108	if (ctx->flags & IORING_SETUP_CQE32)
b3659a65 PB	109	ctx->cqe_cached++;
f3b44f92 JA	110	return cqe;
	111	}
	112
aa1df3a3 PB	113	return __io_get_cqe(ctx, overflow);
	114	}
	115
	116	static inline struct io_uring_cqe io_get_cqe(struct io_ring_ctx ctx)
	117	{
	118	return io_get_cqe_overflow(ctx, false);
f3b44f92 JA	119	}
	120
	121	static inline bool __io_fill_cqe_req(struct io_ring_ctx *ctx,
	122	struct io_kiocb *req)
	123	{
	124	struct io_uring_cqe *cqe;
	125
e8c328c3 PB	126	/*
	127	* If we can't get a cq entry, userspace overflowed the
	128	* submission (by quite a lot). Increment the overflow count in
	129	* the ring.
	130	*/
	131	cqe = io_get_cqe(ctx);
	132	if (unlikely(!cqe))
	133	return io_req_cqe_overflow(req);
e0486f3f DY	134
	135	trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
	136	req->cqe.res, req->cqe.flags,
	137	(req->flags & REQ_F_CQE32_INIT) ? req->extra1 : 0,
	138	(req->flags & REQ_F_CQE32_INIT) ? req->extra2 : 0);
	139
e8c328c3 PB	140	memcpy(cqe, &req->cqe, sizeof(*cqe));
	141
	142	if (ctx->flags & IORING_SETUP_CQE32) {
f3b44f92 JA	143	u64 extra1 = 0, extra2 = 0;
	144
	145	if (req->flags & REQ_F_CQE32_INIT) {
	146	extra1 = req->extra1;
	147	extra2 = req->extra2;
	148	}
	149
e8c328c3 PB	150	WRITE_ONCE(cqe->big_cqe[0], extra1);
e8c328c3 PB	151	WRITE_ONCE(cqe->big_cqe[1], extra2);
f3b44f92	152	}
e8c328c3	153	return true;
f3b44f92 JA	154	}
f3b44f92 JA	155
531113bb JA	156	static inline void req_set_fail(struct io_kiocb *req)
	157	{
	158	req->flags \|= REQ_F_FAIL;
	159	if (req->flags & REQ_F_CQE_SKIP) {
	160	req->flags &= ~REQ_F_CQE_SKIP;
	161	req->flags \|= REQ_F_SKIP_LINK_CQES;
	162	}
	163	}
	164
de23077e JA	165	static inline void io_req_set_res(struct io_kiocb *req, s32 res, u32 cflags)
	166	{
	167	req->cqe.res = res;
	168	req->cqe.flags = cflags;
	169	}
	170
99f15d8d JA	171	static inline bool req_has_async_data(struct io_kiocb *req)
	172	{
	173	return req->flags & REQ_F_ASYNC_DATA;
	174	}
	175
531113bb JA	176	static inline void io_put_file(struct file *file)
	177	{
	178	if (file)
	179	fput(file);
	180	}
	181
cd40cae2 JA	182	static inline void io_ring_submit_unlock(struct io_ring_ctx *ctx,
	183	unsigned issue_flags)
	184	{
	185	lockdep_assert_held(&ctx->uring_lock);
	186	if (issue_flags & IO_URING_F_UNLOCKED)
	187	mutex_unlock(&ctx->uring_lock);
	188	}
	189
	190	static inline void io_ring_submit_lock(struct io_ring_ctx *ctx,
	191	unsigned issue_flags)
	192	{
	193	/*
	194	* "Normal" inline submissions always hold the uring_lock, since we
	195	* grab it from the system call. Same is true for the SQPOLL offload.
	196	* The only exception is when we've detached the request and issue it
	197	* from an async worker thread, grab the lock for that case.
	198	*/
	199	if (issue_flags & IO_URING_F_UNLOCKED)
	200	mutex_lock(&ctx->uring_lock);
	201	lockdep_assert_held(&ctx->uring_lock);
	202	}
	203
f9ead18c JA	204	static inline void io_commit_cqring(struct io_ring_ctx *ctx)
	205	{
	206	/* order cqe stores with ring update */
	207	smp_store_release(&ctx->rings->cq.tail, ctx->cached_cq_tail);
	208	}
	209
fc86f9d3 PB	210	/* requires smb_mb() prior, see wq_has_sleeper() */
fc86f9d3 PB	211	static inline void __io_cqring_wake(struct io_ring_ctx *ctx)
f3b44f92 JA	212	{
f3b44f92 JA	213	/*
44648532 JA	214	* Trigger waitqueue handler on all waiters on our waitqueue. This
	215	* won't necessarily wake up all the tasks, io_should_wake() will make
	216	* that decision.
	217	*
	218	* Pass in EPOLLIN\|EPOLL_URING_WAKE as the poll wakeup key. The latter
	219	* set in the mask so that if we recurse back into our own poll
	220	* waitqueue handlers, we know we have a dependency between eventfd or
	221	* epoll and should terminate multishot poll at that point.
f3b44f92	222	*/
fc86f9d3	223	if (waitqueue_active(&ctx->cq_wait))
44648532 JA	224	__wake_up(&ctx->cq_wait, TASK_NORMAL, 0,
44648532 JA	225	poll_to_key(EPOLL_URING_WAKE \| EPOLLIN));
f3b44f92 JA	226	}
f3b44f92 JA	227
fc86f9d3 PB	228	static inline void io_cqring_wake(struct io_ring_ctx *ctx)
	229	{
	230	smp_mb();
	231	__io_cqring_wake(ctx);
	232	}
	233
17437f31 JA	234	static inline bool io_sqring_full(struct io_ring_ctx *ctx)
	235	{
	236	struct io_rings *r = ctx->rings;
	237
	238	return READ_ONCE(r->sq.tail) - ctx->cached_sq_head == ctx->sq_entries;
	239	}
	240
	241	static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
	242	{
	243	struct io_rings *rings = ctx->rings;
	244
	245	/* make sure SQ entry isn't read before tail */
	246	return smp_load_acquire(&rings->sq.tail) - ctx->cached_sq_head;
	247	}
	248
c0e0d6ba	249	static inline int io_run_task_work(void)
17437f31	250	{
46a525e1 JA	251	if (task_work_pending(current)) {
	252	if (test_thread_flag(TIF_NOTIFY_SIGNAL))
	253	clear_notify_signal();
17437f31	254	__set_current_state(TASK_RUNNING);
46a525e1	255	task_work_run();
c0e0d6ba	256	return 1;
17437f31 JA	257	}
17437f31 JA	258
c0e0d6ba DY	259	return 0;
	260	}
	261
dac6a0ea JA	262	static inline bool io_task_work_pending(struct io_ring_ctx *ctx)
	263	{
	264	return test_thread_flag(TIF_NOTIFY_SIGNAL) \|\|
	265	!wq_list_empty(&ctx->work_llist);
	266	}
	267
c0e0d6ba DY	268	static inline int io_run_task_work_ctx(struct io_ring_ctx *ctx)
	269	{
	270	int ret = 0;
	271	int ret2;
	272
	273	if (ctx->flags & IORING_SETUP_DEFER_TASKRUN)
	274	ret = io_run_local_work(ctx);
	275
	276	/* want to run this after in case more is added */
	277	ret2 = io_run_task_work();
	278
	279	/* Try propagate error in favour of if tasks were run,
	280	* but still make sure to run them if requested
	281	*/
	282	if (ret >= 0)
	283	ret += ret2;
	284
	285	return ret;
17437f31 JA	286	}
17437f31 JA	287
44f87745 PB	288	static inline int io_run_local_work_locked(struct io_ring_ctx *ctx)
44f87745 PB	289	{
b3026767 DY	290	bool locked;
	291	int ret;
	292
44f87745 PB	293	if (llist_empty(&ctx->work_llist))
44f87745 PB	294	return 0;
b3026767 DY	295
	296	locked = true;
	297	ret = __io_run_local_work(ctx, &locked);
	298	/* shouldn't happen! */
	299	if (WARN_ON_ONCE(!locked))
	300	mutex_lock(&ctx->uring_lock);
	301	return ret;
44f87745 PB	302	}
44f87745 PB	303
aa1e90f6 PB	304	static inline void io_tw_lock(struct io_ring_ctx ctx, bool locked)
	305	{
	306	if (!*locked) {
	307	mutex_lock(&ctx->uring_lock);
	308	*locked = true;
	309	}
	310	}
	311
9da070b1 PB	312	/*
	313	* Don't complete immediately but use deferred completion infrastructure.
	314	* Protected by ->uring_lock and can only be used either with
	315	* IO_URING_F_COMPLETE_DEFER or inside a tw handler holding the mutex.
	316	*/
	317	static inline void io_req_complete_defer(struct io_kiocb *req)
	318	__must_hold(&req->ctx->uring_lock)
aa1e90f6 PB	319	{
	320	struct io_submit_state *state = &req->ctx->submit_state;
	321
9da070b1 PB	322	lockdep_assert_held(&req->ctx->uring_lock);
9da070b1 PB	323
aa1e90f6 PB	324	wq_list_add_tail(&req->comp_list, &state->compl_reqs);
	325	}
	326
46929b08 PB	327	static inline void io_commit_cqring_flush(struct io_ring_ctx *ctx)
	328	{
	329	if (unlikely(ctx->off_timeout_used \|\| ctx->drain_active \|\| ctx->has_evfd))
	330	__io_commit_cqring_flush(ctx);
	331	}
	332
e70cb608 PB	333	/* must to be called somewhat shortly after putting a request */
	334	static inline void io_put_task(struct task_struct *task, int nr)
	335	{
	336	if (likely(task == current))
	337	task->io_uring->cached_refs += nr;
	338	else
	339	__io_put_task(task, nr);
	340	}
	341
63809137 PB	342	static inline void io_get_task_refs(int nr)
	343	{
	344	struct io_uring_task *tctx = current->io_uring;
	345
	346	tctx->cached_refs -= nr;
	347	if (unlikely(tctx->cached_refs < 0))
	348	io_task_refs_refill(tctx);
	349	}
	350
bd1a3783 PB	351	static inline bool io_req_cache_empty(struct io_ring_ctx *ctx)
	352	{
	353	return !ctx->submit_state.free_list.next;
	354	}
	355
	356	static inline bool io_alloc_req_refill(struct io_ring_ctx *ctx)
	357	{
	358	if (unlikely(io_req_cache_empty(ctx)))
	359	return __io_alloc_req_refill(ctx);
	360	return true;
	361	}
	362
	363	static inline struct io_kiocb io_alloc_req(struct io_ring_ctx ctx)
	364	{
	365	struct io_wq_work_node *node;
	366
	367	node = wq_stack_extract(&ctx->submit_state.free_list);
	368	return container_of(node, struct io_kiocb, comp_list);
	369	}
	370
76de6749 PB	371	static inline bool io_allowed_run_tw(struct io_ring_ctx *ctx)
76de6749 PB	372	{
6567506b PB	373	return likely(!(ctx->flags & IORING_SETUP_DEFER_TASKRUN) \|\|
6567506b PB	374	ctx->submitter_task == current);
76de6749 PB	375	}
76de6749 PB	376
833b5dff PB	377	static inline void io_req_queue_tw_complete(struct io_kiocb *req, s32 res)
	378	{
	379	io_req_set_res(req, res, 0);
	380	req->io_task_work.func = io_req_task_complete;
	381	io_req_task_work_add(req);
	382	}
	383
de23077e	384	#endif