[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_defer_failed(struct io_kiocb *req, s32 res);
void io_req_complete_post(struct io_kiocb *req, unsigned issue_flags);
bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
		     bool allow_overflow);
bool io_aux_cqe(struct io_ring_ctx *ctx, bool defer, u64 user_data, s32 res, u32 cflags,
		bool allow_overflow);
void __io_commit_cqring_flush(struct io_ring_ctx *ctx);

static inline void io_req_complete_post_tw(struct io_kiocb *req, bool *locked)
{
	unsigned flags = *locked ? 0 : IO_URING_F_UNLOCKED;

	io_req_complete_post(req, flags);
}

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);
973fc83f	33	void io_req_defer_failed(struct io_kiocb *req, s32 res);
1bec951c	34	void io_req_complete_post(struct io_kiocb *req, unsigned issue_flags);
52120f0f DY	35	bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags,
52120f0f DY	36	bool allow_overflow);
9b8c5475 DY	37	bool io_aux_cqe(struct io_ring_ctx *ctx, bool defer, u64 user_data, s32 res, u32 cflags,
9b8c5475 DY	38	bool allow_overflow);
9046c641 PB	39	void __io_commit_cqring_flush(struct io_ring_ctx *ctx);
9046c641 PB	40
1bec951c PB	41	static inline void io_req_complete_post_tw(struct io_kiocb req, bool locked)
	42	{
	43	unsigned flags = *locked ? 0 : IO_URING_F_UNLOCKED;
	44
	45	io_req_complete_post(req, flags);
	46	}
	47
9046c641 PB	48	struct page *io_pin_pages(unsigned long ubuf, unsigned long len, int npages);
	49
	50	struct file io_file_get_normal(struct io_kiocb req, int fd);
	51	struct file io_file_get_fixed(struct io_kiocb req, int fd,
	52	unsigned issue_flags);
	53
f6b543fd JA	54	static inline bool io_req_ffs_set(struct io_kiocb *req)
	55	{
	56	return req->flags & REQ_F_FIXED_FILE;
	57	}
	58
e52d2e58	59	void __io_req_task_work_add(struct io_kiocb *req, bool allow_local);
9046c641 PB	60	bool io_is_uring_fops(struct file *file);
9046c641 PB	61	bool io_alloc_async_data(struct io_kiocb *req);
9046c641 PB	62	void io_req_task_queue(struct io_kiocb *req);
	63	void io_queue_iowq(struct io_kiocb req, bool dont_use);
	64	void io_req_task_complete(struct io_kiocb req, bool locked);
	65	void io_req_task_queue_fail(struct io_kiocb *req, int ret);
	66	void io_req_task_submit(struct io_kiocb req, bool locked);
	67	void tctx_task_work(struct callback_head *cb);
	68	__cold void io_uring_cancel_generic(bool cancel_all, struct io_sq_data *sqd);
	69	int io_uring_alloc_task_context(struct task_struct *task,
	70	struct io_ring_ctx *ctx);
	71
	72	int io_poll_issue(struct io_kiocb req, bool locked);
	73	int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr);
	74	int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin);
	75	void io_free_batch_list(struct io_ring_ctx ctx, struct io_wq_work_node node);
	76	int io_req_prep_async(struct io_kiocb *req);
	77
	78	struct io_wq_work io_wq_free_work(struct io_wq_work work);
	79	void io_wq_submit_work(struct io_wq_work *work);
	80
	81	void io_free_req(struct io_kiocb *req);
	82	void io_queue_next(struct io_kiocb *req);
e70cb608	83	void __io_put_task(struct task_struct *task, int nr);
63809137	84	void io_task_refs_refill(struct io_uring_task *tctx);
bd1a3783	85	bool __io_alloc_req_refill(struct io_ring_ctx *ctx);
9046c641 PB	86
	87	bool io_match_task_safe(struct io_kiocb head, struct task_struct task,
	88	bool cancel_all);
	89
e52d2e58 PB	90	static inline void io_req_task_work_add(struct io_kiocb *req)
	91	{
	92	__io_req_task_work_add(req, true);
	93	}
	94
9046c641 PB	95	#define io_for_each_link(pos, head) \
9046c641 PB	96	for (pos = (head); pos; pos = pos->link)
f3b44f92	97
25399321 PB	98	static inline void io_cq_lock(struct io_ring_ctx *ctx)
	99	__acquires(ctx->completion_lock)
	100	{
	101	spin_lock(&ctx->completion_lock);
	102	}
	103
	104	void io_cq_unlock_post(struct io_ring_ctx *ctx);
	105
aa1df3a3 PB	106	static inline struct io_uring_cqe io_get_cqe_overflow(struct io_ring_ctx ctx,
aa1df3a3 PB	107	bool overflow)
f3b44f92 JA	108	{
	109	if (likely(ctx->cqe_cached < ctx->cqe_sentinel)) {
	110	struct io_uring_cqe *cqe = ctx->cqe_cached;
	111
f3b44f92 JA	112	ctx->cached_cq_tail++;
f3b44f92 JA	113	ctx->cqe_cached++;
b3659a65 PB	114	if (ctx->flags & IORING_SETUP_CQE32)
b3659a65 PB	115	ctx->cqe_cached++;
f3b44f92 JA	116	return cqe;
	117	}
	118
aa1df3a3 PB	119	return __io_get_cqe(ctx, overflow);
	120	}
	121
	122	static inline struct io_uring_cqe io_get_cqe(struct io_ring_ctx ctx)
	123	{
	124	return io_get_cqe_overflow(ctx, false);
f3b44f92 JA	125	}
	126
	127	static inline bool __io_fill_cqe_req(struct io_ring_ctx *ctx,
	128	struct io_kiocb *req)
	129	{
	130	struct io_uring_cqe *cqe;
	131
e8c328c3 PB	132	/*
	133	* If we can't get a cq entry, userspace overflowed the
	134	* submission (by quite a lot). Increment the overflow count in
	135	* the ring.
	136	*/
	137	cqe = io_get_cqe(ctx);
	138	if (unlikely(!cqe))
	139	return io_req_cqe_overflow(req);
e0486f3f DY	140
	141	trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
	142	req->cqe.res, req->cqe.flags,
	143	(req->flags & REQ_F_CQE32_INIT) ? req->extra1 : 0,
	144	(req->flags & REQ_F_CQE32_INIT) ? req->extra2 : 0);
	145
e8c328c3 PB	146	memcpy(cqe, &req->cqe, sizeof(*cqe));
	147
	148	if (ctx->flags & IORING_SETUP_CQE32) {
f3b44f92 JA	149	u64 extra1 = 0, extra2 = 0;
	150
	151	if (req->flags & REQ_F_CQE32_INIT) {
	152	extra1 = req->extra1;
	153	extra2 = req->extra2;
	154	}
	155
e8c328c3 PB	156	WRITE_ONCE(cqe->big_cqe[0], extra1);
e8c328c3 PB	157	WRITE_ONCE(cqe->big_cqe[1], extra2);
f3b44f92	158	}
e8c328c3	159	return true;
f3b44f92 JA	160	}
f3b44f92 JA	161
531113bb JA	162	static inline void req_set_fail(struct io_kiocb *req)
	163	{
	164	req->flags \|= REQ_F_FAIL;
	165	if (req->flags & REQ_F_CQE_SKIP) {
	166	req->flags &= ~REQ_F_CQE_SKIP;
	167	req->flags \|= REQ_F_SKIP_LINK_CQES;
	168	}
	169	}
	170
de23077e JA	171	static inline void io_req_set_res(struct io_kiocb *req, s32 res, u32 cflags)
	172	{
	173	req->cqe.res = res;
	174	req->cqe.flags = cflags;
	175	}
	176
99f15d8d JA	177	static inline bool req_has_async_data(struct io_kiocb *req)
	178	{
	179	return req->flags & REQ_F_ASYNC_DATA;
	180	}
	181
531113bb JA	182	static inline void io_put_file(struct file *file)
	183	{
	184	if (file)
	185	fput(file);
	186	}
	187
cd40cae2 JA	188	static inline void io_ring_submit_unlock(struct io_ring_ctx *ctx,
	189	unsigned issue_flags)
	190	{
	191	lockdep_assert_held(&ctx->uring_lock);
	192	if (issue_flags & IO_URING_F_UNLOCKED)
	193	mutex_unlock(&ctx->uring_lock);
	194	}
	195
	196	static inline void io_ring_submit_lock(struct io_ring_ctx *ctx,
	197	unsigned issue_flags)
	198	{
	199	/*
	200	* "Normal" inline submissions always hold the uring_lock, since we
	201	* grab it from the system call. Same is true for the SQPOLL offload.
	202	* The only exception is when we've detached the request and issue it
	203	* from an async worker thread, grab the lock for that case.
	204	*/
	205	if (issue_flags & IO_URING_F_UNLOCKED)
	206	mutex_lock(&ctx->uring_lock);
	207	lockdep_assert_held(&ctx->uring_lock);
	208	}
	209
f9ead18c JA	210	static inline void io_commit_cqring(struct io_ring_ctx *ctx)
	211	{
	212	/* order cqe stores with ring update */
	213	smp_store_release(&ctx->rings->cq.tail, ctx->cached_cq_tail);
	214	}
	215
fc86f9d3 PB	216	/* requires smb_mb() prior, see wq_has_sleeper() */
fc86f9d3 PB	217	static inline void __io_cqring_wake(struct io_ring_ctx *ctx)
f3b44f92 JA	218	{
f3b44f92 JA	219	/*
44648532 JA	220	* Trigger waitqueue handler on all waiters on our waitqueue. This
	221	* won't necessarily wake up all the tasks, io_should_wake() will make
	222	* that decision.
	223	*
	224	* Pass in EPOLLIN\|EPOLL_URING_WAKE as the poll wakeup key. The latter
	225	* set in the mask so that if we recurse back into our own poll
	226	* waitqueue handlers, we know we have a dependency between eventfd or
	227	* epoll and should terminate multishot poll at that point.
f3b44f92	228	*/
fc86f9d3	229	if (waitqueue_active(&ctx->cq_wait))
44648532 JA	230	__wake_up(&ctx->cq_wait, TASK_NORMAL, 0,
44648532 JA	231	poll_to_key(EPOLL_URING_WAKE \| EPOLLIN));
f3b44f92 JA	232	}
f3b44f92 JA	233
fc86f9d3 PB	234	static inline void io_cqring_wake(struct io_ring_ctx *ctx)
	235	{
	236	smp_mb();
	237	__io_cqring_wake(ctx);
	238	}
	239
17437f31 JA	240	static inline bool io_sqring_full(struct io_ring_ctx *ctx)
	241	{
	242	struct io_rings *r = ctx->rings;
	243
	244	return READ_ONCE(r->sq.tail) - ctx->cached_sq_head == ctx->sq_entries;
	245	}
	246
	247	static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
	248	{
	249	struct io_rings *rings = ctx->rings;
	250
	251	/* make sure SQ entry isn't read before tail */
	252	return smp_load_acquire(&rings->sq.tail) - ctx->cached_sq_head;
	253	}
	254
c0e0d6ba	255	static inline int io_run_task_work(void)
17437f31	256	{
46a525e1 JA	257	if (task_work_pending(current)) {
	258	if (test_thread_flag(TIF_NOTIFY_SIGNAL))
	259	clear_notify_signal();
17437f31	260	__set_current_state(TASK_RUNNING);
46a525e1	261	task_work_run();
c0e0d6ba	262	return 1;
17437f31 JA	263	}
17437f31 JA	264
c0e0d6ba DY	265	return 0;
	266	}
	267
dac6a0ea JA	268	static inline bool io_task_work_pending(struct io_ring_ctx *ctx)
	269	{
	270	return test_thread_flag(TIF_NOTIFY_SIGNAL) \|\|
	271	!wq_list_empty(&ctx->work_llist);
	272	}
	273
c0e0d6ba DY	274	static inline int io_run_task_work_ctx(struct io_ring_ctx *ctx)
	275	{
	276	int ret = 0;
	277	int ret2;
	278
	279	if (ctx->flags & IORING_SETUP_DEFER_TASKRUN)
	280	ret = io_run_local_work(ctx);
	281
	282	/* want to run this after in case more is added */
	283	ret2 = io_run_task_work();
	284
	285	/* Try propagate error in favour of if tasks were run,
	286	* but still make sure to run them if requested
	287	*/
	288	if (ret >= 0)
	289	ret += ret2;
	290
	291	return ret;
17437f31 JA	292	}
17437f31 JA	293
44f87745 PB	294	static inline int io_run_local_work_locked(struct io_ring_ctx *ctx)
44f87745 PB	295	{
b3026767 DY	296	bool locked;
	297	int ret;
	298
44f87745 PB	299	if (llist_empty(&ctx->work_llist))
44f87745 PB	300	return 0;
b3026767 DY	301
	302	locked = true;
	303	ret = __io_run_local_work(ctx, &locked);
	304	/* shouldn't happen! */
	305	if (WARN_ON_ONCE(!locked))
	306	mutex_lock(&ctx->uring_lock);
	307	return ret;
44f87745 PB	308	}
44f87745 PB	309
aa1e90f6 PB	310	static inline void io_tw_lock(struct io_ring_ctx ctx, bool locked)
	311	{
	312	if (!*locked) {
	313	mutex_lock(&ctx->uring_lock);
	314	*locked = true;
	315	}
	316	}
	317
9da070b1 PB	318	/*
	319	* Don't complete immediately but use deferred completion infrastructure.
	320	* Protected by ->uring_lock and can only be used either with
	321	* IO_URING_F_COMPLETE_DEFER or inside a tw handler holding the mutex.
	322	*/
	323	static inline void io_req_complete_defer(struct io_kiocb *req)
	324	__must_hold(&req->ctx->uring_lock)
aa1e90f6 PB	325	{
	326	struct io_submit_state *state = &req->ctx->submit_state;
	327
9da070b1 PB	328	lockdep_assert_held(&req->ctx->uring_lock);
9da070b1 PB	329
aa1e90f6 PB	330	wq_list_add_tail(&req->comp_list, &state->compl_reqs);
	331	}
	332
46929b08 PB	333	static inline void io_commit_cqring_flush(struct io_ring_ctx *ctx)
	334	{
	335	if (unlikely(ctx->off_timeout_used \|\| ctx->drain_active \|\| ctx->has_evfd))
	336	__io_commit_cqring_flush(ctx);
	337	}
	338
e70cb608 PB	339	/* must to be called somewhat shortly after putting a request */
	340	static inline void io_put_task(struct task_struct *task, int nr)
	341	{
	342	if (likely(task == current))
	343	task->io_uring->cached_refs += nr;
	344	else
	345	__io_put_task(task, nr);
	346	}
	347
63809137 PB	348	static inline void io_get_task_refs(int nr)
	349	{
	350	struct io_uring_task *tctx = current->io_uring;
	351
	352	tctx->cached_refs -= nr;
	353	if (unlikely(tctx->cached_refs < 0))
	354	io_task_refs_refill(tctx);
	355	}
	356
bd1a3783 PB	357	static inline bool io_req_cache_empty(struct io_ring_ctx *ctx)
	358	{
	359	return !ctx->submit_state.free_list.next;
	360	}
	361
	362	static inline bool io_alloc_req_refill(struct io_ring_ctx *ctx)
	363	{
	364	if (unlikely(io_req_cache_empty(ctx)))
	365	return __io_alloc_req_refill(ctx);
	366	return true;
	367	}
	368
	369	static inline struct io_kiocb io_alloc_req(struct io_ring_ctx ctx)
	370	{
	371	struct io_wq_work_node *node;
	372
	373	node = wq_stack_extract(&ctx->submit_state.free_list);
	374	return container_of(node, struct io_kiocb, comp_list);
	375	}
	376
76de6749 PB	377	static inline bool io_allowed_run_tw(struct io_ring_ctx *ctx)
76de6749 PB	378	{
6567506b PB	379	return likely(!(ctx->flags & IORING_SETUP_DEFER_TASKRUN) \|\|
6567506b PB	380	ctx->submitter_task == current);
76de6749 PB	381	}
76de6749 PB	382
833b5dff PB	383	static inline void io_req_queue_tw_complete(struct io_kiocb *req, s32 res)
	384	{
	385	io_req_set_res(req, res, 0);
	386	req->io_task_work.func = io_req_task_complete;
	387	io_req_task_work_add(req);
	388	}
	389
de23077e	390	#endif