[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 "io_uring_types.h"

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

enum {
	IOU_OK			= 0,
	IOU_ISSUE_SKIP_COMPLETE	= -EIOCBQUEUED,
};

struct io_uring_cqe *__io_get_cqe(struct io_ring_ctx *ctx);
bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data, s32 res,
			      u32 cflags, u64 extra1, u64 extra2);

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

		if (ctx->flags & IORING_SETUP_CQE32) {
			unsigned int off = ctx->cqe_cached - ctx->rings->cqes;

			cqe += off;
		}

		ctx->cached_cq_tail++;
		ctx->cqe_cached++;
		return cqe;
	}

	return __io_get_cqe(ctx);
}

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

	if (!(ctx->flags & IORING_SETUP_CQE32)) {
		trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
					req->cqe.res, req->cqe.flags, 0, 0);

		/*
		 * 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 (likely(cqe)) {
			memcpy(cqe, &req->cqe, sizeof(*cqe));
			return true;
		}

		return io_cqring_event_overflow(ctx, req->cqe.user_data,
						req->cqe.res, req->cqe.flags,
						0, 0);
	} else {
		u64 extra1 = 0, extra2 = 0;

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

		trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
					req->cqe.res, req->cqe.flags, extra1, extra2);

		/*
		 * 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 (likely(cqe)) {
			memcpy(cqe, &req->cqe, sizeof(struct io_uring_cqe));
			WRITE_ONCE(cqe->big_cqe[0], extra1);
			WRITE_ONCE(cqe->big_cqe[1], extra2);
			return true;
		}

		return io_cqring_event_overflow(ctx, req->cqe.user_data,
				req->cqe.res, req->cqe.flags,
				extra1, extra2);
	}
}

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);
}

static inline void io_cqring_wake(struct io_ring_ctx *ctx)
{
	/*
	 * wake_up_all() may seem excessive, but io_wake_function() and
	 * io_should_wake() handle the termination of the loop and only
	 * wake as many waiters as we need to.
	 */
	if (wq_has_sleeper(&ctx->cq_wait))
		wake_up_all(&ctx->cq_wait);
}

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 bool io_run_task_work(void)
{
	if (test_thread_flag(TIF_NOTIFY_SIGNAL) || task_work_pending(current)) {
		__set_current_state(TASK_RUNNING);
		clear_notify_signal();
		if (task_work_pending(current))
			task_work_run();
		return true;
	}

	return false;
}

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

static inline void io_req_add_compl_list(struct io_kiocb *req)
{
	struct io_submit_state *state = &req->ctx->submit_state;

	if (!(req->flags & REQ_F_CQE_SKIP))
		state->flush_cqes = true;
	wq_list_add_tail(&req->comp_list, &state->compl_reqs);
}

int io_run_task_work_sig(void);
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);
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);
void io_cqring_ev_posted(struct io_ring_ctx *ctx);
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);

bool io_is_uring_fops(struct file *file);
bool io_alloc_async_data(struct io_kiocb *req);
void io_req_task_work_add(struct io_kiocb *req);
void io_req_task_prio_work_add(struct io_kiocb *req);
void io_req_tw_post_queue(struct io_kiocb *req, s32 res, u32 cflags);
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);

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

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

#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>
de23077e JA	6	#include "io_uring_types.h"
de23077e JA	7
f3b44f92 JA	8	#ifndef CREATE_TRACE_POINTS
	9	#include <trace/events/io_uring.h>
	10	#endif
	11
97b388d7 JA	12	enum {
	13	IOU_OK = 0,
	14	IOU_ISSUE_SKIP_COMPLETE = -EIOCBQUEUED,
	15	};
	16
faf88dde	17	struct io_uring_cqe __io_get_cqe(struct io_ring_ctx ctx);
f3b44f92 JA	18	bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data, s32 res,
	19	u32 cflags, u64 extra1, u64 extra2);
	20
f3b44f92 JA	21	static inline struct io_uring_cqe io_get_cqe(struct io_ring_ctx ctx)
	22	{
	23	if (likely(ctx->cqe_cached < ctx->cqe_sentinel)) {
	24	struct io_uring_cqe *cqe = ctx->cqe_cached;
	25
	26	if (ctx->flags & IORING_SETUP_CQE32) {
	27	unsigned int off = ctx->cqe_cached - ctx->rings->cqes;
	28
	29	cqe += off;
	30	}
	31
	32	ctx->cached_cq_tail++;
	33	ctx->cqe_cached++;
	34	return cqe;
	35	}
	36
	37	return __io_get_cqe(ctx);
	38	}
	39
	40	static inline bool __io_fill_cqe_req(struct io_ring_ctx *ctx,
	41	struct io_kiocb *req)
	42	{
	43	struct io_uring_cqe *cqe;
	44
	45	if (!(ctx->flags & IORING_SETUP_CQE32)) {
	46	trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
	47	req->cqe.res, req->cqe.flags, 0, 0);
	48
	49	/*
	50	* If we can't get a cq entry, userspace overflowed the
	51	* submission (by quite a lot). Increment the overflow count in
	52	* the ring.
	53	*/
	54	cqe = io_get_cqe(ctx);
	55	if (likely(cqe)) {
	56	memcpy(cqe, &req->cqe, sizeof(*cqe));
	57	return true;
	58	}
	59
	60	return io_cqring_event_overflow(ctx, req->cqe.user_data,
	61	req->cqe.res, req->cqe.flags,
	62	0, 0);
	63	} else {
	64	u64 extra1 = 0, extra2 = 0;
	65
	66	if (req->flags & REQ_F_CQE32_INIT) {
	67	extra1 = req->extra1;
	68	extra2 = req->extra2;
	69	}
	70
	71	trace_io_uring_complete(req->ctx, req, req->cqe.user_data,
	72	req->cqe.res, req->cqe.flags, extra1, extra2);
	73
	74	/*
	75	* If we can't get a cq entry, userspace overflowed the
	76	* submission (by quite a lot). Increment the overflow count in
	77	* the ring.
	78	*/
	79	cqe = io_get_cqe(ctx);
	80	if (likely(cqe)) {
	81	memcpy(cqe, &req->cqe, sizeof(struct io_uring_cqe));
	82	WRITE_ONCE(cqe->big_cqe[0], extra1);
	83	WRITE_ONCE(cqe->big_cqe[1], extra2);
	84	return true;
85	}
86
87	return io_cqring_event_overflow(ctx, req->cqe.user_data,
88	req->cqe.res, req->cqe.flags,
89	extra1, extra2);
90	}
91	}
92
531113bb JA	93	static inline void req_set_fail(struct io_kiocb *req)
	94	{
	95	req->flags \|= REQ_F_FAIL;
	96	if (req->flags & REQ_F_CQE_SKIP) {
	97	req->flags &= ~REQ_F_CQE_SKIP;
	98	req->flags \|= REQ_F_SKIP_LINK_CQES;
	99	}
	100	}
	101
de23077e JA	102	static inline void io_req_set_res(struct io_kiocb *req, s32 res, u32 cflags)
	103	{
	104	req->cqe.res = res;
	105	req->cqe.flags = cflags;
	106	}
	107
99f15d8d JA	108	static inline bool req_has_async_data(struct io_kiocb *req)
	109	{
	110	return req->flags & REQ_F_ASYNC_DATA;
	111	}
	112
531113bb JA	113	static inline void io_put_file(struct file *file)
	114	{
	115	if (file)
	116	fput(file);
	117	}
	118
cd40cae2 JA	119	static inline void io_ring_submit_unlock(struct io_ring_ctx *ctx,
	120	unsigned issue_flags)
	121	{
	122	lockdep_assert_held(&ctx->uring_lock);
	123	if (issue_flags & IO_URING_F_UNLOCKED)
	124	mutex_unlock(&ctx->uring_lock);
	125	}
	126
	127	static inline void io_ring_submit_lock(struct io_ring_ctx *ctx,
	128	unsigned issue_flags)
	129	{
	130	/*
	131	* "Normal" inline submissions always hold the uring_lock, since we
	132	* grab it from the system call. Same is true for the SQPOLL offload.
	133	* The only exception is when we've detached the request and issue it
	134	* from an async worker thread, grab the lock for that case.
	135	*/
	136	if (issue_flags & IO_URING_F_UNLOCKED)
	137	mutex_lock(&ctx->uring_lock);
	138	lockdep_assert_held(&ctx->uring_lock);
	139	}
	140
f9ead18c JA	141	static inline void io_commit_cqring(struct io_ring_ctx *ctx)
	142	{
	143	/* order cqe stores with ring update */
	144	smp_store_release(&ctx->rings->cq.tail, ctx->cached_cq_tail);
	145	}
	146
f3b44f92 JA	147	static inline void io_cqring_wake(struct io_ring_ctx *ctx)
	148	{
	149	/*
	150	* wake_up_all() may seem excessive, but io_wake_function() and
	151	* io_should_wake() handle the termination of the loop and only
	152	* wake as many waiters as we need to.
	153	*/
	154	if (wq_has_sleeper(&ctx->cq_wait))
	155	wake_up_all(&ctx->cq_wait);
	156	}
	157
17437f31 JA	158	static inline bool io_sqring_full(struct io_ring_ctx *ctx)
	159	{
	160	struct io_rings *r = ctx->rings;
	161
	162	return READ_ONCE(r->sq.tail) - ctx->cached_sq_head == ctx->sq_entries;
	163	}
	164
	165	static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
	166	{
	167	struct io_rings *rings = ctx->rings;
	168
	169	/* make sure SQ entry isn't read before tail */
	170	return smp_load_acquire(&rings->sq.tail) - ctx->cached_sq_head;
	171	}
	172
	173	static inline bool io_run_task_work(void)
	174	{
	175	if (test_thread_flag(TIF_NOTIFY_SIGNAL) \|\| task_work_pending(current)) {
	176	__set_current_state(TASK_RUNNING);
	177	clear_notify_signal();
	178	if (task_work_pending(current))
	179	task_work_run();
	180	return true;
	181	}
	182
	183	return false;
	184	}
	185
aa1e90f6 PB	186	static inline void io_tw_lock(struct io_ring_ctx ctx, bool locked)
	187	{
	188	if (!*locked) {
	189	mutex_lock(&ctx->uring_lock);
	190	*locked = true;
	191	}
	192	}
	193
	194	static inline void io_req_add_compl_list(struct io_kiocb *req)
	195	{
	196	struct io_submit_state *state = &req->ctx->submit_state;
	197
	198	if (!(req->flags & REQ_F_CQE_SKIP))
	199	state->flush_cqes = true;
	200	wq_list_add_tail(&req->comp_list, &state->compl_reqs);
	201	}
	202
73572984	203	int io_run_task_work_sig(void);
329061d3	204	void io_req_complete_failed(struct io_kiocb *req, s32 res);
99f15d8d	205	void __io_req_complete(struct io_kiocb *req, unsigned issue_flags);
59915143 JA	206	void io_req_complete_post(struct io_kiocb *req);
59915143 JA	207	void __io_req_complete_post(struct io_kiocb *req);
d245bca6	208	bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags);
f9ead18c	209	void io_cqring_ev_posted(struct io_ring_ctx *ctx);
f3b44f92	210	void __io_commit_cqring_flush(struct io_ring_ctx *ctx);
f9ead18c	211
3b77495a	212	struct page *io_pin_pages(unsigned long ubuf, unsigned long len, int npages);
329061d3	213
531113bb JA	214	struct file io_file_get_normal(struct io_kiocb req, int fd);
	215	struct file io_file_get_fixed(struct io_kiocb req, int fd,
	216	unsigned issue_flags);
cd40cae2	217
cd40cae2	218	bool io_is_uring_fops(struct file *file);
99f15d8d JA	219	bool io_alloc_async_data(struct io_kiocb *req);
99f15d8d JA	220	void io_req_task_work_add(struct io_kiocb *req);
f3b44f92	221	void io_req_task_prio_work_add(struct io_kiocb *req);
59915143	222	void io_req_tw_post_queue(struct io_kiocb *req, s32 res, u32 cflags);
f3b44f92 JA	223	void io_req_task_queue(struct io_kiocb *req);
f3b44f92 JA	224	void io_queue_iowq(struct io_kiocb req, bool dont_use);
59915143 JA	225	void io_req_task_complete(struct io_kiocb req, bool locked);
59915143 JA	226	void io_req_task_queue_fail(struct io_kiocb *req, int ret);
329061d3	227	void io_req_task_submit(struct io_kiocb req, bool locked);
c9f06aa7	228	void tctx_task_work(struct callback_head *cb);
17437f31 JA	229	__cold void io_uring_cancel_generic(bool cancel_all, struct io_sq_data *sqd);
	230	int io_uring_alloc_task_context(struct task_struct *task,
	231	struct io_ring_ctx *ctx);
	232
329061d3	233	int io_poll_issue(struct io_kiocb req, bool locked);
17437f31 JA	234	int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr);
17437f31 JA	235	int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin);
f3b44f92 JA	236	void io_free_batch_list(struct io_ring_ctx ctx, struct io_wq_work_node node);
f3b44f92 JA	237	int io_req_prep_async(struct io_kiocb *req);
59915143	238
c9f06aa7 JA	239	struct io_wq_work io_wq_free_work(struct io_wq_work work);
	240	void io_wq_submit_work(struct io_wq_work *work);
	241
59915143 JA	242	void io_free_req(struct io_kiocb *req);
	243	void io_queue_next(struct io_kiocb *req);
	244
329061d3 JA	245	bool io_match_task_safe(struct io_kiocb head, struct task_struct task,
	246	bool cancel_all);
	247
59915143 JA	248	#define io_for_each_link(pos, head) \
59915143 JA	249	for (pos = (head); pos; pos = pos->link)
531113bb	250
de23077e	251	#endif