io_uring/io_uring.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef IOU_CORE_H
#define IOU_CORE_H

#include <linux/errno.h>
#include <linux/file.h>
#include <linux/lockdep.h>
#include <linux/resume_user_mode.h>
#include <linux/poll.h>
#include <linux/io_uring_types.h>
#include <uapi/linux/eventpoll.h>
#include "alloc_cache.h"
#include "io-wq.h"
#include "slist.h"
#include "tw.h"
#include "opdef.h"

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

struct io_rings_layout {
	/* size of CQ + headers + SQ offset array */
	size_t rings_size;
	size_t sq_size;

	size_t sq_array_offset;
};

struct io_ctx_config {
	struct io_uring_params p;
	struct io_rings_layout layout;
	struct io_uring_params __user *uptr;
};

#define IORING_FEAT_FLAGS (IORING_FEAT_SINGLE_MMAP |\
			IORING_FEAT_NODROP |\
			IORING_FEAT_SUBMIT_STABLE |\
			IORING_FEAT_RW_CUR_POS |\
			IORING_FEAT_CUR_PERSONALITY |\
			IORING_FEAT_FAST_POLL |\
			IORING_FEAT_POLL_32BITS |\
			IORING_FEAT_SQPOLL_NONFIXED |\
			IORING_FEAT_EXT_ARG |\
			IORING_FEAT_NATIVE_WORKERS |\
			IORING_FEAT_RSRC_TAGS |\
			IORING_FEAT_CQE_SKIP |\
			IORING_FEAT_LINKED_FILE |\
			IORING_FEAT_REG_REG_RING |\
			IORING_FEAT_RECVSEND_BUNDLE |\
			IORING_FEAT_MIN_TIMEOUT |\
			IORING_FEAT_RW_ATTR |\
			IORING_FEAT_NO_IOWAIT)

#define IORING_SETUP_FLAGS (IORING_SETUP_IOPOLL |\
			IORING_SETUP_SQPOLL |\
			IORING_SETUP_SQ_AFF |\
			IORING_SETUP_CQSIZE |\
			IORING_SETUP_CLAMP |\
			IORING_SETUP_ATTACH_WQ |\
			IORING_SETUP_R_DISABLED |\
			IORING_SETUP_SUBMIT_ALL |\
			IORING_SETUP_COOP_TASKRUN |\
			IORING_SETUP_TASKRUN_FLAG |\
			IORING_SETUP_SQE128 |\
			IORING_SETUP_CQE32 |\
			IORING_SETUP_SINGLE_ISSUER |\
			IORING_SETUP_DEFER_TASKRUN |\
			IORING_SETUP_NO_MMAP |\
			IORING_SETUP_REGISTERED_FD_ONLY |\
			IORING_SETUP_NO_SQARRAY |\
			IORING_SETUP_HYBRID_IOPOLL |\
			IORING_SETUP_CQE_MIXED |\
			IORING_SETUP_SQE_MIXED |\
			IORING_SETUP_SQ_REWIND)

#define IORING_ENTER_FLAGS (IORING_ENTER_GETEVENTS |\
			IORING_ENTER_SQ_WAKEUP |\
			IORING_ENTER_SQ_WAIT |\
			IORING_ENTER_EXT_ARG |\
			IORING_ENTER_REGISTERED_RING |\
			IORING_ENTER_ABS_TIMER |\
			IORING_ENTER_EXT_ARG_REG |\
			IORING_ENTER_NO_IOWAIT)


#define SQE_VALID_FLAGS (IOSQE_FIXED_FILE |\
			IOSQE_IO_DRAIN |\
			IOSQE_IO_LINK |\
			IOSQE_IO_HARDLINK |\
			IOSQE_ASYNC |\
			IOSQE_BUFFER_SELECT |\
			IOSQE_CQE_SKIP_SUCCESS)

#define IO_REQ_LINK_FLAGS (REQ_F_LINK | REQ_F_HARDLINK)

/*
 * Complaint timeout for io_uring cancelation exits, and for io-wq exit
 * worker waiting.
 */
#define IO_URING_EXIT_WAIT_MAX	(HZ * 60 * 5)

enum {
	IOU_COMPLETE		= 0,

	IOU_ISSUE_SKIP_COMPLETE	= -EIOCBQUEUED,

	/*
	 * The request has more work to do and should be retried. io_uring will
	 * attempt to wait on the file for eligible opcodes, but otherwise
	 * it'll be handed to iowq for blocking execution. It works for normal
	 * requests as well as for the multi shot mode.
	 */
	IOU_RETRY		= -EAGAIN,

	/*
	 * Requeue the task_work to restart operations on this request. The
	 * actual value isn't important, should just be not an otherwise
	 * valid error code, yet less than -MAX_ERRNO and valid internally.
	 */
	IOU_REQUEUE		= -3072,
};

struct io_defer_entry {
	struct list_head	list;
	struct io_kiocb		*req;
};

struct io_wait_queue {
	struct wait_queue_entry wq;
	struct io_ring_ctx *ctx;
	unsigned cq_tail;
	unsigned cq_min_tail;
	unsigned nr_timeouts;
	int hit_timeout;
	ktime_t min_timeout;
	ktime_t timeout;
	struct hrtimer t;

#ifdef CONFIG_NET_RX_BUSY_POLL
	ktime_t napi_busy_poll_dt;
	bool napi_prefer_busy_poll;
#endif
};

static inline struct io_rings *io_get_rings(struct io_ring_ctx *ctx)
{
	return rcu_dereference_check(ctx->rings_rcu,
			lockdep_is_held(&ctx->uring_lock) ||
			lockdep_is_held(&ctx->completion_lock));
}

static inline bool io_should_wake(struct io_wait_queue *iowq)
{
	struct io_ring_ctx *ctx = iowq->ctx;
	struct io_rings *rings;
	int dist;

	guard(rcu)();
	rings = io_get_rings(ctx);

	/*
	 * Wake up if we have enough events, or if a timeout occurred since we
	 * started waiting. For timeouts, we always want to return to userspace,
	 * regardless of event count.
	 */
	dist = READ_ONCE(rings->cq.tail) - (int) iowq->cq_tail;
	return dist >= 0 || atomic_read(&ctx->cq_timeouts) != iowq->nr_timeouts;
}

#define IORING_MAX_ENTRIES	32768
#define IORING_MAX_CQ_ENTRIES	(2 * IORING_MAX_ENTRIES)

int io_prepare_config(struct io_ctx_config *config);

bool io_cqe_cache_refill(struct io_ring_ctx *ctx, bool overflow, bool cqe32);
void io_req_defer_failed(struct io_kiocb *req, s32 res);
bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags);
void io_add_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags);
bool io_req_post_cqe(struct io_kiocb *req, s32 res, u32 cflags);
bool io_req_post_cqe32(struct io_kiocb *req, struct io_uring_cqe src_cqe[2]);
void __io_commit_cqring_flush(struct io_ring_ctx *ctx);

unsigned io_linked_nr(struct io_kiocb *req);
void io_req_track_inflight(struct io_kiocb *req);
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);
struct file *io_uring_ctx_get_file(unsigned int fd, bool registered);

void io_req_task_queue(struct io_kiocb *req);
void io_req_task_complete(struct io_tw_req tw_req, io_tw_token_t tw);
void io_req_task_queue_fail(struct io_kiocb *req, int ret);
void io_req_task_submit(struct io_tw_req tw_req, io_tw_token_t tw);
__cold void io_uring_drop_tctx_refs(struct task_struct *task);

int io_ring_add_registered_file(struct io_uring_task *tctx, struct file *file,
				     int start, int end);
void io_queue_iowq(struct io_kiocb *req);

int io_poll_issue(struct io_kiocb *req, io_tw_token_t tw);
int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr);
int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin);
__cold void io_iopoll_try_reap_events(struct io_ring_ctx *ctx);
void __io_submit_flush_completions(struct io_ring_ctx *ctx);

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_task_refs_refill(struct io_uring_task *tctx);
bool __io_alloc_req_refill(struct io_ring_ctx *ctx);

void io_activate_pollwq(struct io_ring_ctx *ctx);
void io_restriction_clone(struct io_restriction *dst, struct io_restriction *src);
void io_poison_req(struct io_kiocb *req);

static inline void io_lockdep_assert_cq_locked(struct io_ring_ctx *ctx)
{
#if defined(CONFIG_PROVE_LOCKING)
	lockdep_assert(in_task());

	if (ctx->flags & IORING_SETUP_DEFER_TASKRUN)
		lockdep_assert_held(&ctx->uring_lock);

	if (ctx->flags & IORING_SETUP_IOPOLL) {
		lockdep_assert_held(&ctx->uring_lock);
	} else if (!(ctx->int_flags & IO_RING_F_TASK_COMPLETE)) {
		lockdep_assert_held(&ctx->completion_lock);
	} else if (ctx->submitter_task) {
		/*
		 * ->submitter_task may be NULL and we can still post a CQE,
		 * if the ring has been setup with IORING_SETUP_R_DISABLED.
		 * Not from an SQE, as those cannot be submitted, but via
		 * updating tagged resources.
		 */
		if (!percpu_ref_is_dying(&ctx->refs))
			lockdep_assert(current == ctx->submitter_task);
	}
#endif
}

static inline bool io_is_compat(struct io_ring_ctx *ctx)
{
	return IS_ENABLED(CONFIG_COMPAT) && unlikely(ctx->int_flags & IO_RING_F_COMPAT);
}

static inline void io_submit_flush_completions(struct io_ring_ctx *ctx)
{
	if (!wq_list_empty(&ctx->submit_state.compl_reqs) ||
	    ctx->submit_state.cq_flush)
		__io_submit_flush_completions(ctx);
}

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

static inline bool io_get_cqe_overflow(struct io_ring_ctx *ctx,
					struct io_uring_cqe **ret,
					bool overflow, bool cqe32)
{
	io_lockdep_assert_cq_locked(ctx);

	if (unlikely(ctx->cqe_sentinel - ctx->cqe_cached < (cqe32 + 1))) {
		if (unlikely(!io_cqe_cache_refill(ctx, overflow, cqe32)))
			return false;
	}
	*ret = ctx->cqe_cached;
	ctx->cached_cq_tail++;
	ctx->cqe_cached++;
	if (ctx->flags & IORING_SETUP_CQE32) {
		ctx->cqe_cached++;
	} else if (cqe32 && ctx->flags & IORING_SETUP_CQE_MIXED) {
		ctx->cqe_cached++;
		ctx->cached_cq_tail++;
	}
	WARN_ON_ONCE(ctx->cqe_cached > ctx->cqe_sentinel);
	return true;
}

static inline bool io_get_cqe(struct io_ring_ctx *ctx, struct io_uring_cqe **ret,
				bool cqe32)
{
	return io_get_cqe_overflow(ctx, ret, false, cqe32);
}

static inline bool io_defer_get_uncommited_cqe(struct io_ring_ctx *ctx,
					       struct io_uring_cqe **cqe_ret)
{
	io_lockdep_assert_cq_locked(ctx);

	ctx->submit_state.cq_flush = true;
	return io_get_cqe(ctx, cqe_ret, ctx->flags & IORING_SETUP_CQE_MIXED);
}

static __always_inline bool io_fill_cqe_req(struct io_ring_ctx *ctx,
					    struct io_kiocb *req)
{
	bool is_cqe32 = req->cqe.flags & IORING_CQE_F_32;
	struct io_uring_cqe *cqe;

	/*
	 * If we can't get a cq entry, userspace overflowed the submission
	 * (by quite a lot).
	 */
	if (unlikely(!io_get_cqe(ctx, &cqe, is_cqe32)))
		return false;

	memcpy(cqe, &req->cqe, sizeof(*cqe));
	if (ctx->flags & IORING_SETUP_CQE32 || is_cqe32) {
		memcpy(cqe->big_cqe, &req->big_cqe, sizeof(*cqe));
		memset(&req->big_cqe, 0, sizeof(req->big_cqe));
	}

	if (trace_io_uring_complete_enabled())
		trace_call__io_uring_complete(req->ctx, req, cqe);
	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 u32 ctx_cqe32_flags(struct io_ring_ctx *ctx)
{
	if (ctx->flags & IORING_SETUP_CQE_MIXED)
		return IORING_CQE_F_32;
	return 0;
}

static inline void io_req_set_res32(struct io_kiocb *req, s32 res, u32 cflags,
				    __u64 extra1, __u64 extra2)
{
	req->cqe.res = res;
	req->cqe.flags = cflags | ctx_cqe32_flags(req->ctx);
	req->big_cqe.extra1 = extra1;
	req->big_cqe.extra2 = extra2;
}

static inline void *io_uring_alloc_async_data(struct io_alloc_cache *cache,
					      struct io_kiocb *req)
{
	if (cache) {
		req->async_data = io_cache_alloc(cache, GFP_KERNEL);
	} else {
		const struct io_issue_def *def = &io_issue_defs[req->opcode];

		WARN_ON_ONCE(!def->async_size);
		req->async_data = kmalloc(def->async_size, GFP_KERNEL);
	}
	if (req->async_data)
		req->flags |= REQ_F_ASYNC_DATA;
	return req->async_data;
}

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

static inline void io_req_async_data_clear(struct io_kiocb *req,
					   io_req_flags_t extra_flags)
{
	req->flags &= ~(REQ_F_ASYNC_DATA|extra_flags);
	req->async_data = NULL;
}

static inline void io_req_async_data_free(struct io_kiocb *req)
{
	kfree(req->async_data);
	io_req_async_data_clear(req, 0);
}

static inline void io_put_file(struct io_kiocb *req)
{
	if (!(req->flags & REQ_F_FIXED_FILE) && req->file)
		fput(req->file);
}

static inline void io_ring_submit_unlock(struct io_ring_ctx *ctx,
					 unsigned issue_flags)
{
	lockdep_assert_held(&ctx->uring_lock);
	if (unlikely(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 (unlikely(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_wq_wake(struct wait_queue_head *wq)
{
	/*
	 *
	 * 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 (wq_has_sleeper(wq))
		__wake_up(wq, TASK_NORMAL, 0, poll_to_key(EPOLL_URING_WAKE | EPOLLIN));
}

static inline void io_poll_wq_wake(struct io_ring_ctx *ctx)
{
	__io_wq_wake(&ctx->poll_wq);
}

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.
	 */

	__io_wq_wake(&ctx->cq_wait);
}

static inline bool __io_sqring_full(struct io_ring_ctx *ctx)
{
	struct io_rings *r = io_get_rings(ctx);

	/*
	 * SQPOLL must use the actual sqring head, as using the cached_sq_head
	 * is race prone if the SQPOLL thread has grabbed entries but not yet
	 * committed them to the ring. For !SQPOLL, this doesn't matter, but
	 * since this helper is just used for SQPOLL sqring waits (or POLLOUT),
	 * just read the actual sqring head unconditionally.
	 */
	return READ_ONCE(r->sq.tail) - READ_ONCE(r->sq.head) == ctx->sq_entries;
}

static inline bool io_sqring_full(struct io_ring_ctx *ctx)
{
	guard(rcu)();
	return __io_sqring_full(ctx);
}

static inline unsigned int __io_sqring_entries(struct io_ring_ctx *ctx)
{
	struct io_rings *rings = io_get_rings(ctx);
	unsigned int entries;

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

static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
{
	guard(rcu)();
	return __io_sqring_entries(ctx);
}

/*
 * 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);
}

#define SHOULD_FLUSH_MASK	(IO_RING_F_OFF_TIMEOUT_USED | \
				 IO_RING_F_HAS_EVFD | IO_RING_F_POLL_ACTIVATED)

static inline void io_commit_cqring_flush(struct io_ring_ctx *ctx)
{
	if (unlikely(data_race(ctx->int_flags) & SHOULD_FLUSH_MASK))
		__io_commit_cqring_flush(ctx);
}

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;
}

extern struct kmem_cache *req_cachep;

static inline struct io_kiocb *io_extract_req(struct io_ring_ctx *ctx)
{
	struct io_kiocb *req;

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

static inline bool io_alloc_req(struct io_ring_ctx *ctx, struct io_kiocb **req)
{
	if (unlikely(io_req_cache_empty(ctx))) {
		if (!__io_alloc_req_refill(ctx))
			return false;
	}
	*req = io_extract_req(ctx);
	return true;
}

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);
}

static inline bool io_file_can_poll(struct io_kiocb *req)
{
	if (req->flags & REQ_F_CAN_POLL)
		return true;
	if (req->file && file_can_poll(req->file)) {
		req->flags |= REQ_F_CAN_POLL;
		return true;
	}
	return false;
}

static inline bool io_is_uring_cmd(const struct io_kiocb *req)
{
	return req->opcode == IORING_OP_URING_CMD ||
	       req->opcode == IORING_OP_URING_CMD128;
}

static inline ktime_t io_get_time(struct io_ring_ctx *ctx)
{
	if (ctx->clockid == CLOCK_MONOTONIC)
		return ktime_get();

	return ktime_get_with_offset(ctx->clock_offset);
}

enum {
	IO_CHECK_CQ_OVERFLOW_BIT,
	IO_CHECK_CQ_DROPPED_BIT,
};

static inline bool io_has_work(struct io_ring_ctx *ctx)
{
	return test_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq) ||
	       io_local_work_pending(ctx);
}
#endif