[linux-2.6-block.git] / include / linux / io_uring_types.h

#ifndef IO_URING_TYPES_H
#define IO_URING_TYPES_H

#include <linux/blkdev.h>
#include <linux/task_work.h>
#include <linux/bitmap.h>
#include <linux/llist.h>
#include <uapi/linux/io_uring.h>

struct io_wq_work_node {
	struct io_wq_work_node *next;
};

struct io_wq_work_list {
	struct io_wq_work_node *first;
	struct io_wq_work_node *last;
};

struct io_wq_work {
	struct io_wq_work_node list;
	unsigned flags;
	/* place it here instead of io_kiocb as it fills padding and saves 4B */
	int cancel_seq;
};

struct io_fixed_file {
	/* file * with additional FFS_* flags */
	unsigned long file_ptr;
};

struct io_file_table {
	struct io_fixed_file *files;
	unsigned long *bitmap;
	unsigned int alloc_hint;
};

struct io_hash_bucket {
	spinlock_t		lock;
	struct hlist_head	list;
} ____cacheline_aligned_in_smp;

struct io_hash_table {
	struct io_hash_bucket	*hbs;
	unsigned		hash_bits;
};

/*
 * Arbitrary limit, can be raised if need be
 */
#define IO_RINGFD_REG_MAX 16

struct io_uring_task {
	/* submission side */
	int				cached_refs;
	const struct io_ring_ctx 	*last;
	struct io_wq			*io_wq;
	struct file			*registered_rings[IO_RINGFD_REG_MAX];

	struct xarray			xa;
	struct wait_queue_head		wait;
	atomic_t			in_cancel;
	atomic_t			inflight_tracked;
	struct percpu_counter		inflight;

	struct { /* task_work */
		struct llist_head	task_list;
		struct callback_head	task_work;
	} ____cacheline_aligned_in_smp;
};

struct io_uring {
	u32 head ____cacheline_aligned_in_smp;
	u32 tail ____cacheline_aligned_in_smp;
};

/*
 * This data is shared with the application through the mmap at offsets
 * IORING_OFF_SQ_RING and IORING_OFF_CQ_RING.
 *
 * The offsets to the member fields are published through struct
 * io_sqring_offsets when calling io_uring_setup.
 */
struct io_rings {
	/*
	 * Head and tail offsets into the ring; the offsets need to be
	 * masked to get valid indices.
	 *
	 * The kernel controls head of the sq ring and the tail of the cq ring,
	 * and the application controls tail of the sq ring and the head of the
	 * cq ring.
	 */
	struct io_uring		sq, cq;
	/*
	 * Bitmasks to apply to head and tail offsets (constant, equals
	 * ring_entries - 1)
	 */
	u32			sq_ring_mask, cq_ring_mask;
	/* Ring sizes (constant, power of 2) */
	u32			sq_ring_entries, cq_ring_entries;
	/*
	 * Number of invalid entries dropped by the kernel due to
	 * invalid index stored in array
	 *
	 * Written by the kernel, shouldn't be modified by the
	 * application (i.e. get number of "new events" by comparing to
	 * cached value).
	 *
	 * After a new SQ head value was read by the application this
	 * counter includes all submissions that were dropped reaching
	 * the new SQ head (and possibly more).
	 */
	u32			sq_dropped;
	/*
	 * Runtime SQ flags
	 *
	 * Written by the kernel, shouldn't be modified by the
	 * application.
	 *
	 * The application needs a full memory barrier before checking
	 * for IORING_SQ_NEED_WAKEUP after updating the sq tail.
	 */
	atomic_t		sq_flags;
	/*
	 * Runtime CQ flags
	 *
	 * Written by the application, shouldn't be modified by the
	 * kernel.
	 */
	u32			cq_flags;
	/*
	 * Number of completion events lost because the queue was full;
	 * this should be avoided by the application by making sure
	 * there are not more requests pending than there is space in
	 * the completion queue.
	 *
	 * Written by the kernel, shouldn't be modified by the
	 * application (i.e. get number of "new events" by comparing to
	 * cached value).
	 *
	 * As completion events come in out of order this counter is not
	 * ordered with any other data.
	 */
	u32			cq_overflow;
	/*
	 * Ring buffer of completion events.
	 *
	 * The kernel writes completion events fresh every time they are
	 * produced, so the application is allowed to modify pending
	 * entries.
	 */
	struct io_uring_cqe	cqes[] ____cacheline_aligned_in_smp;
};

struct io_restriction {
	DECLARE_BITMAP(register_op, IORING_REGISTER_LAST);
	DECLARE_BITMAP(sqe_op, IORING_OP_LAST);
	u8 sqe_flags_allowed;
	u8 sqe_flags_required;
	bool registered;
};

struct io_submit_link {
	struct io_kiocb		*head;
	struct io_kiocb		*last;
};

struct io_submit_state {
	/* inline/task_work completion list, under ->uring_lock */
	struct io_wq_work_node	free_list;
	/* batch completion logic */
	struct io_wq_work_list	compl_reqs;
	struct io_submit_link	link;

	bool			plug_started;
	bool			need_plug;
	unsigned short		submit_nr;
	unsigned int		cqes_count;
	struct blk_plug		plug;
	struct io_uring_cqe	cqes[16];
};

struct io_ev_fd {
	struct eventfd_ctx	*cq_ev_fd;
	unsigned int		eventfd_async: 1;
	struct rcu_head		rcu;
	atomic_t		refs;
	atomic_t		ops;
};

struct io_alloc_cache {
	struct io_wq_work_node	list;
	unsigned int		nr_cached;
	unsigned int		max_cached;
	size_t			elem_size;
};

struct io_ring_ctx {
	/* const or read-mostly hot data */
	struct {
		unsigned int		flags;
		unsigned int		drain_next: 1;
		unsigned int		restricted: 1;
		unsigned int		off_timeout_used: 1;
		unsigned int		drain_active: 1;
		unsigned int		has_evfd: 1;
		/* all CQEs should be posted only by the submitter task */
		unsigned int		task_complete: 1;
		unsigned int		syscall_iopoll: 1;
		unsigned int		poll_activated: 1;
		unsigned int		drain_disabled: 1;
		unsigned int		compat: 1;

		enum task_work_notify_mode	notify_method;

		/*
		 * If IORING_SETUP_NO_MMAP is used, then the below holds
		 * the gup'ed pages for the two rings, and the sqes.
		 */
		unsigned short		n_ring_pages;
		unsigned short		n_sqe_pages;
		struct page		**ring_pages;
		struct page		**sqe_pages;

		struct io_rings			*rings;
		struct task_struct		*submitter_task;
		struct percpu_ref		refs;
	} ____cacheline_aligned_in_smp;

	/* submission data */
	struct {
		struct mutex		uring_lock;

		/*
		 * Ring buffer of indices into array of io_uring_sqe, which is
		 * mmapped by the application using the IORING_OFF_SQES offset.
		 *
		 * This indirection could e.g. be used to assign fixed
		 * io_uring_sqe entries to operations and only submit them to
		 * the queue when needed.
		 *
		 * The kernel modifies neither the indices array nor the entries
		 * array.
		 */
		u32			*sq_array;
		struct io_uring_sqe	*sq_sqes;
		unsigned		cached_sq_head;
		unsigned		sq_entries;

		/*
		 * Fixed resources fast path, should be accessed only under
		 * uring_lock, and updated through io_uring_register(2)
		 */
		struct io_rsrc_node	*rsrc_node;
		atomic_t		cancel_seq;
		struct io_file_table	file_table;
		unsigned		nr_user_files;
		unsigned		nr_user_bufs;
		struct io_mapped_ubuf	**user_bufs;

		struct io_submit_state	submit_state;

		struct io_buffer_list	*io_bl;
		struct xarray		io_bl_xa;
		struct list_head	io_buffers_cache;

		struct io_hash_table	cancel_table_locked;
		struct list_head	cq_overflow_list;
		struct io_alloc_cache	apoll_cache;
		struct io_alloc_cache	netmsg_cache;
	} ____cacheline_aligned_in_smp;

	/* IRQ completion list, under ->completion_lock */
	struct io_wq_work_list	locked_free_list;
	unsigned int		locked_free_nr;

	const struct cred	*sq_creds;	/* cred used for __io_sq_thread() */
	struct io_sq_data	*sq_data;	/* if using sq thread polling */

	struct wait_queue_head	sqo_sq_wait;
	struct list_head	sqd_list;

	unsigned long		check_cq;

	unsigned int		file_alloc_start;
	unsigned int		file_alloc_end;

	struct xarray		personalities;
	u32			pers_next;

	struct {
		/*
		 * We cache a range of free CQEs we can use, once exhausted it
		 * should go through a slower range setup, see __io_get_cqe()
		 */
		struct io_uring_cqe	*cqe_cached;
		struct io_uring_cqe	*cqe_sentinel;

		unsigned		cached_cq_tail;
		unsigned		cq_entries;
		struct io_ev_fd	__rcu	*io_ev_fd;
		struct wait_queue_head	cq_wait;
		unsigned		cq_extra;
	} ____cacheline_aligned_in_smp;

	struct {
		spinlock_t		completion_lock;

		bool			poll_multi_queue;
		atomic_t		cq_wait_nr;

		/*
		 * ->iopoll_list is protected by the ctx->uring_lock for
		 * io_uring instances that don't use IORING_SETUP_SQPOLL.
		 * For SQPOLL, only the single threaded io_sq_thread() will
		 * manipulate the list, hence no extra locking is needed there.
		 */
		struct io_wq_work_list	iopoll_list;
		struct io_hash_table	cancel_table;

		struct llist_head	work_llist;

		struct list_head	io_buffers_comp;
	} ____cacheline_aligned_in_smp;

	/* timeouts */
	struct {
		spinlock_t		timeout_lock;
		atomic_t		cq_timeouts;
		struct list_head	timeout_list;
		struct list_head	ltimeout_list;
		unsigned		cq_last_tm_flush;
	} ____cacheline_aligned_in_smp;

	/* Keep this last, we don't need it for the fast path */
	struct wait_queue_head		poll_wq;
	struct io_restriction		restrictions;

	/* slow path rsrc auxilary data, used by update/register */
	struct io_mapped_ubuf		*dummy_ubuf;
	struct io_rsrc_data		*file_data;
	struct io_rsrc_data		*buf_data;

	/* protected by ->uring_lock */
	struct list_head		rsrc_ref_list;
	struct io_alloc_cache		rsrc_node_cache;
	struct wait_queue_head		rsrc_quiesce_wq;
	unsigned			rsrc_quiesce;

	struct list_head		io_buffers_pages;

	#if defined(CONFIG_UNIX)
		struct socket		*ring_sock;
	#endif
	/* hashed buffered write serialization */
	struct io_wq_hash		*hash_map;

	/* Only used for accounting purposes */
	struct user_struct		*user;
	struct mm_struct		*mm_account;

	/* ctx exit and cancelation */
	struct llist_head		fallback_llist;
	struct delayed_work		fallback_work;
	struct work_struct		exit_work;
	struct list_head		tctx_list;
	struct completion		ref_comp;

	/* io-wq management, e.g. thread count */
	u32				iowq_limits[2];
	bool				iowq_limits_set;

	struct callback_head		poll_wq_task_work;
	struct list_head		defer_list;
	unsigned			sq_thread_idle;
	/* protected by ->completion_lock */
	unsigned			evfd_last_cq_tail;
};

struct io_tw_state {
	/* ->uring_lock is taken, callbacks can use io_tw_lock to lock it */
	bool locked;
};

enum {
	REQ_F_FIXED_FILE_BIT	= IOSQE_FIXED_FILE_BIT,
	REQ_F_IO_DRAIN_BIT	= IOSQE_IO_DRAIN_BIT,
	REQ_F_LINK_BIT		= IOSQE_IO_LINK_BIT,
	REQ_F_HARDLINK_BIT	= IOSQE_IO_HARDLINK_BIT,
	REQ_F_FORCE_ASYNC_BIT	= IOSQE_ASYNC_BIT,
	REQ_F_BUFFER_SELECT_BIT	= IOSQE_BUFFER_SELECT_BIT,
	REQ_F_CQE_SKIP_BIT	= IOSQE_CQE_SKIP_SUCCESS_BIT,

	/* first byte is taken by user flags, shift it to not overlap */
	REQ_F_FAIL_BIT		= 8,
	REQ_F_INFLIGHT_BIT,
	REQ_F_CUR_POS_BIT,
	REQ_F_NOWAIT_BIT,
	REQ_F_LINK_TIMEOUT_BIT,
	REQ_F_NEED_CLEANUP_BIT,
	REQ_F_POLLED_BIT,
	REQ_F_BUFFER_SELECTED_BIT,
	REQ_F_BUFFER_RING_BIT,
	REQ_F_REISSUE_BIT,
	REQ_F_CREDS_BIT,
	REQ_F_REFCOUNT_BIT,
	REQ_F_ARM_LTIMEOUT_BIT,
	REQ_F_ASYNC_DATA_BIT,
	REQ_F_SKIP_LINK_CQES_BIT,
	REQ_F_SINGLE_POLL_BIT,
	REQ_F_DOUBLE_POLL_BIT,
	REQ_F_PARTIAL_IO_BIT,
	REQ_F_APOLL_MULTISHOT_BIT,
	REQ_F_CLEAR_POLLIN_BIT,
	REQ_F_HASH_LOCKED_BIT,
	/* keep async read/write and isreg together and in order */
	REQ_F_SUPPORT_NOWAIT_BIT,
	REQ_F_ISREG_BIT,

	/* not a real bit, just to check we're not overflowing the space */
	__REQ_F_LAST_BIT,
};

enum {
	/* ctx owns file */
	REQ_F_FIXED_FILE	= BIT(REQ_F_FIXED_FILE_BIT),
	/* drain existing IO first */
	REQ_F_IO_DRAIN		= BIT(REQ_F_IO_DRAIN_BIT),
	/* linked sqes */
	REQ_F_LINK		= BIT(REQ_F_LINK_BIT),
	/* doesn't sever on completion < 0 */
	REQ_F_HARDLINK		= BIT(REQ_F_HARDLINK_BIT),
	/* IOSQE_ASYNC */
	REQ_F_FORCE_ASYNC	= BIT(REQ_F_FORCE_ASYNC_BIT),
	/* IOSQE_BUFFER_SELECT */
	REQ_F_BUFFER_SELECT	= BIT(REQ_F_BUFFER_SELECT_BIT),
	/* IOSQE_CQE_SKIP_SUCCESS */
	REQ_F_CQE_SKIP		= BIT(REQ_F_CQE_SKIP_BIT),

	/* fail rest of links */
	REQ_F_FAIL		= BIT(REQ_F_FAIL_BIT),
	/* on inflight list, should be cancelled and waited on exit reliably */
	REQ_F_INFLIGHT		= BIT(REQ_F_INFLIGHT_BIT),
	/* read/write uses file position */
	REQ_F_CUR_POS		= BIT(REQ_F_CUR_POS_BIT),
	/* must not punt to workers */
	REQ_F_NOWAIT		= BIT(REQ_F_NOWAIT_BIT),
	/* has or had linked timeout */
	REQ_F_LINK_TIMEOUT	= BIT(REQ_F_LINK_TIMEOUT_BIT),
	/* needs cleanup */
	REQ_F_NEED_CLEANUP	= BIT(REQ_F_NEED_CLEANUP_BIT),
	/* already went through poll handler */
	REQ_F_POLLED		= BIT(REQ_F_POLLED_BIT),
	/* buffer already selected */
	REQ_F_BUFFER_SELECTED	= BIT(REQ_F_BUFFER_SELECTED_BIT),
	/* buffer selected from ring, needs commit */
	REQ_F_BUFFER_RING	= BIT(REQ_F_BUFFER_RING_BIT),
	/* caller should reissue async */
	REQ_F_REISSUE		= BIT(REQ_F_REISSUE_BIT),
	/* supports async reads/writes */
	REQ_F_SUPPORT_NOWAIT	= BIT(REQ_F_SUPPORT_NOWAIT_BIT),
	/* regular file */
	REQ_F_ISREG		= BIT(REQ_F_ISREG_BIT),
	/* has creds assigned */
	REQ_F_CREDS		= BIT(REQ_F_CREDS_BIT),
	/* skip refcounting if not set */
	REQ_F_REFCOUNT		= BIT(REQ_F_REFCOUNT_BIT),
	/* there is a linked timeout that has to be armed */
	REQ_F_ARM_LTIMEOUT	= BIT(REQ_F_ARM_LTIMEOUT_BIT),
	/* ->async_data allocated */
	REQ_F_ASYNC_DATA	= BIT(REQ_F_ASYNC_DATA_BIT),
	/* don't post CQEs while failing linked requests */
	REQ_F_SKIP_LINK_CQES	= BIT(REQ_F_SKIP_LINK_CQES_BIT),
	/* single poll may be active */
	REQ_F_SINGLE_POLL	= BIT(REQ_F_SINGLE_POLL_BIT),
	/* double poll may active */
	REQ_F_DOUBLE_POLL	= BIT(REQ_F_DOUBLE_POLL_BIT),
	/* request has already done partial IO */
	REQ_F_PARTIAL_IO	= BIT(REQ_F_PARTIAL_IO_BIT),
	/* fast poll multishot mode */
	REQ_F_APOLL_MULTISHOT	= BIT(REQ_F_APOLL_MULTISHOT_BIT),
	/* recvmsg special flag, clear EPOLLIN */
	REQ_F_CLEAR_POLLIN	= BIT(REQ_F_CLEAR_POLLIN_BIT),
	/* hashed into ->cancel_hash_locked, protected by ->uring_lock */
	REQ_F_HASH_LOCKED	= BIT(REQ_F_HASH_LOCKED_BIT),
};

typedef void (*io_req_tw_func_t)(struct io_kiocb *req, struct io_tw_state *ts);

struct io_task_work {
	struct llist_node		node;
	io_req_tw_func_t		func;
};

struct io_cqe {
	__u64	user_data;
	__s32	res;
	/* fd initially, then cflags for completion */
	union {
		__u32	flags;
		int	fd;
	};
};

/*
 * Each request type overlays its private data structure on top of this one.
 * They must not exceed this one in size.
 */
struct io_cmd_data {
	struct file		*file;
	/* each command gets 56 bytes of data */
	__u8			data[56];
};

static inline void io_kiocb_cmd_sz_check(size_t cmd_sz)
{
	BUILD_BUG_ON(cmd_sz > sizeof(struct io_cmd_data));
}
#define io_kiocb_to_cmd(req, cmd_type) ( \
	io_kiocb_cmd_sz_check(sizeof(cmd_type)) , \
	((cmd_type *)&(req)->cmd) \
)
#define cmd_to_io_kiocb(ptr)	((struct io_kiocb *) ptr)

struct io_kiocb {
	union {
		/*
		 * NOTE! Each of the io_kiocb union members has the file pointer
		 * as the first entry in their struct definition. So you can
		 * access the file pointer through any of the sub-structs,
		 * or directly as just 'file' in this struct.
		 */
		struct file		*file;
		struct io_cmd_data	cmd;
	};

	u8				opcode;
	/* polled IO has completed */
	u8				iopoll_completed;
	/*
	 * Can be either a fixed buffer index, or used with provided buffers.
	 * For the latter, before issue it points to the buffer group ID,
	 * and after selection it points to the buffer ID itself.
	 */
	u16				buf_index;
	unsigned int			flags;

	struct io_cqe			cqe;

	struct io_ring_ctx		*ctx;
	struct task_struct		*task;

	struct io_rsrc_node		*rsrc_node;

	union {
		/* store used ubuf, so we can prevent reloading */
		struct io_mapped_ubuf	*imu;

		/* stores selected buf, valid IFF REQ_F_BUFFER_SELECTED is set */
		struct io_buffer	*kbuf;

		/*
		 * stores buffer ID for ring provided buffers, valid IFF
		 * REQ_F_BUFFER_RING is set.
		 */
		struct io_buffer_list	*buf_list;
	};

	union {
		/* used by request caches, completion batching and iopoll */
		struct io_wq_work_node	comp_list;
		/* cache ->apoll->events */
		__poll_t apoll_events;
	};
	atomic_t			refs;
	atomic_t			poll_refs;
	struct io_task_work		io_task_work;
	unsigned			nr_tw;
	/* for polled requests, i.e. IORING_OP_POLL_ADD and async armed poll */
	struct hlist_node		hash_node;
	/* internal polling, see IORING_FEAT_FAST_POLL */
	struct async_poll		*apoll;
	/* opcode allocated if it needs to store data for async defer */
	void				*async_data;
	/* linked requests, IFF REQ_F_HARDLINK or REQ_F_LINK are set */
	struct io_kiocb			*link;
	/* custom credentials, valid IFF REQ_F_CREDS is set */
	const struct cred		*creds;
	struct io_wq_work		work;

	struct {
		u64			extra1;
		u64			extra2;
	} big_cqe;
};

struct io_overflow_cqe {
	struct list_head list;
	struct io_uring_cqe cqe;
};

#endif
Commit	Line	Data
e27f928e JA	1	#ifndef IO_URING_TYPES_H
	2	#define IO_URING_TYPES_H
	3
	4	#include <linux/blkdev.h>
	5	#include <linux/task_work.h>
d9b57aa3	6	#include <linux/bitmap.h>
e70cb608	7	#include <linux/llist.h>
d9b57aa3	8	#include <uapi/linux/io_uring.h>
e27f928e	9
ab1c84d8 PB	10	struct io_wq_work_node {
	11	struct io_wq_work_node *next;
	12	};
	13
	14	struct io_wq_work_list {
	15	struct io_wq_work_node *first;
	16	struct io_wq_work_node *last;
	17	};
	18
	19	struct io_wq_work {
	20	struct io_wq_work_node list;
	21	unsigned flags;
	22	/* place it here instead of io_kiocb as it fills padding and saves 4B */
	23	int cancel_seq;
	24	};
	25
	26	struct io_fixed_file {
	27	/* file * with additional FFS_* flags */
	28	unsigned long file_ptr;
	29	};
	30
	31	struct io_file_table {
	32	struct io_fixed_file *files;
	33	unsigned long *bitmap;
	34	unsigned int alloc_hint;
	35	};
e27f928e	36
e6f89be6 PB	37	struct io_hash_bucket {
	38	spinlock_t lock;
	39	struct hlist_head list;
	40	} ____cacheline_aligned_in_smp;
	41
	42	struct io_hash_table {
	43	struct io_hash_bucket *hbs;
	44	unsigned hash_bits;
	45	};
	46
e70cb608 PB	47	/*
	48	* Arbitrary limit, can be raised if need be
	49	*/
	50	#define IO_RINGFD_REG_MAX 16
	51
	52	struct io_uring_task {
	53	/* submission side */
	54	int cached_refs;
	55	const struct io_ring_ctx *last;
	56	struct io_wq *io_wq;
	57	struct file *registered_rings[IO_RINGFD_REG_MAX];
	58
	59	struct xarray xa;
	60	struct wait_queue_head wait;
8d664282	61	atomic_t in_cancel;
e70cb608 PB	62	atomic_t inflight_tracked;
	63	struct percpu_counter inflight;
	64
	65	struct { /* task_work */
	66	struct llist_head task_list;
	67	struct callback_head task_work;
	68	} ____cacheline_aligned_in_smp;
	69	};
	70
e27f928e JA	71	struct io_uring {
	72	u32 head ____cacheline_aligned_in_smp;
	73	u32 tail ____cacheline_aligned_in_smp;
	74	};
	75
	76	/*
	77	* This data is shared with the application through the mmap at offsets
	78	* IORING_OFF_SQ_RING and IORING_OFF_CQ_RING.
	79	*
	80	* The offsets to the member fields are published through struct
	81	* io_sqring_offsets when calling io_uring_setup.
	82	*/
	83	struct io_rings {
	84	/*
	85	* Head and tail offsets into the ring; the offsets need to be
	86	* masked to get valid indices.
	87	*
	88	* The kernel controls head of the sq ring and the tail of the cq ring,
	89	* and the application controls tail of the sq ring and the head of the
	90	* cq ring.
	91	*/
	92	struct io_uring sq, cq;
	93	/*
	94	* Bitmasks to apply to head and tail offsets (constant, equals
	95	* ring_entries - 1)
	96	*/
	97	u32 sq_ring_mask, cq_ring_mask;
	98	/* Ring sizes (constant, power of 2) */
	99	u32 sq_ring_entries, cq_ring_entries;
	100	/*
	101	* Number of invalid entries dropped by the kernel due to
	102	* invalid index stored in array
	103	*
	104	* Written by the kernel, shouldn't be modified by the
	105	* application (i.e. get number of "new events" by comparing to
	106	* cached value).
	107	*
	108	* After a new SQ head value was read by the application this
	109	* counter includes all submissions that were dropped reaching
	110	* the new SQ head (and possibly more).
	111	*/
	112	u32 sq_dropped;
	113	/*
	114	* Runtime SQ flags
	115	*
	116	* Written by the kernel, shouldn't be modified by the
	117	* application.
	118	*
	119	* The application needs a full memory barrier before checking
	120	* for IORING_SQ_NEED_WAKEUP after updating the sq tail.
	121	*/
	122	atomic_t sq_flags;
	123	/*
	124	* Runtime CQ flags
	125	*
	126	* Written by the application, shouldn't be modified by the
	127	* kernel.
	128	*/
	129	u32 cq_flags;
	130	/*
	131	* Number of completion events lost because the queue was full;
	132	* this should be avoided by the application by making sure
	133	* there are not more requests pending than there is space in
	134	* the completion queue.
135	*
136	* Written by the kernel, shouldn't be modified by the
137	* application (i.e. get number of "new events" by comparing to
138	* cached value).
139	*
140	* As completion events come in out of order this counter is not
141	* ordered with any other data.
142	*/
143	u32 cq_overflow;
144	/*
145	* Ring buffer of completion events.
146	*
147	* The kernel writes completion events fresh every time they are
148	* produced, so the application is allowed to modify pending
149	* entries.
150	*/
151	struct io_uring_cqe cqes[] ____cacheline_aligned_in_smp;
152	};
153
154	struct io_restriction {
155	DECLARE_BITMAP(register_op, IORING_REGISTER_LAST);
156	DECLARE_BITMAP(sqe_op, IORING_OP_LAST);
157	u8 sqe_flags_allowed;
158	u8 sqe_flags_required;
159	bool registered;
160	};
161
162	struct io_submit_link {
163	struct io_kiocb *head;
164	struct io_kiocb *last;
165	};
166
167	struct io_submit_state {
168	/* inline/task_work completion list, under ->uring_lock */
169	struct io_wq_work_node free_list;
170	/* batch completion logic */
171	struct io_wq_work_list compl_reqs;
172	struct io_submit_link link;
173
174	bool plug_started;
175	bool need_plug;
e27f928e	176	unsigned short submit_nr;
931147dd	177	unsigned int cqes_count;
e27f928e	178	struct blk_plug plug;
931147dd	179	struct io_uring_cqe cqes[16];
e27f928e JA	180	};
	181
	182	struct io_ev_fd {
	183	struct eventfd_ctx *cq_ev_fd;
	184	unsigned int eventfd_async: 1;
	185	struct rcu_head rcu;
21a091b9 DY	186	atomic_t refs;
21a091b9 DY	187	atomic_t ops;
e27f928e JA	188	};
e27f928e JA	189
9b797a37	190	struct io_alloc_cache {
efba1a9e	191	struct io_wq_work_node list;
9731bc98	192	unsigned int nr_cached;
69bbc6ad	193	unsigned int max_cached;
e1fe7ee8	194	size_t elem_size;
9b797a37 JA	195	};
9b797a37 JA	196
e27f928e JA	197	struct io_ring_ctx {
	198	/* const or read-mostly hot data */
	199	struct {
e27f928e	200	unsigned int flags;
e27f928e JA	201	unsigned int drain_next: 1;
	202	unsigned int restricted: 1;
	203	unsigned int off_timeout_used: 1;
	204	unsigned int drain_active: 1;
e27f928e	205	unsigned int has_evfd: 1;
e6aeb272 PB	206	/* all CQEs should be posted only by the submitter task */
e6aeb272 PB	207	unsigned int task_complete: 1;
632ffe09	208	unsigned int syscall_iopoll: 1;
bca39f39	209	unsigned int poll_activated: 1;
632ffe09 PB	210	unsigned int drain_disabled: 1;
632ffe09 PB	211	unsigned int compat: 1;
dde40322 PB	212
dde40322 PB	213	enum task_work_notify_mode notify_method;
03d89a2d JA	214
	215	/*
	216	* If IORING_SETUP_NO_MMAP is used, then the below holds
	217	* the gup'ed pages for the two rings, and the sqes.
	218	*/
	219	unsigned short n_ring_pages;
	220	unsigned short n_sqe_pages;
	221	struct page **ring_pages;
	222	struct page **sqe_pages;
	223
dde40322 PB	224	struct io_rings *rings;
	225	struct task_struct *submitter_task;
	226	struct percpu_ref refs;
e27f928e JA	227	} ____cacheline_aligned_in_smp;
	228
	229	/* submission data */
	230	struct {
	231	struct mutex uring_lock;
	232
	233	/*
	234	* Ring buffer of indices into array of io_uring_sqe, which is
	235	* mmapped by the application using the IORING_OFF_SQES offset.
	236	*
	237	* This indirection could e.g. be used to assign fixed
	238	* io_uring_sqe entries to operations and only submit them to
	239	* the queue when needed.
	240	*
	241	* The kernel modifies neither the indices array nor the entries
	242	* array.
	243	*/
	244	u32 *sq_array;
	245	struct io_uring_sqe *sq_sqes;
	246	unsigned cached_sq_head;
	247	unsigned sq_entries;
e27f928e JA	248
	249	/*
	250	* Fixed resources fast path, should be accessed only under
	251	* uring_lock, and updated through io_uring_register(2)
	252	*/
	253	struct io_rsrc_node *rsrc_node;
e27f928e JA	254	atomic_t cancel_seq;
	255	struct io_file_table file_table;
	256	unsigned nr_user_files;
	257	unsigned nr_user_bufs;
	258	struct io_mapped_ubuf **user_bufs;
	259
	260	struct io_submit_state submit_state;
	261
	262	struct io_buffer_list *io_bl;
	263	struct xarray io_bl_xa;
	264	struct list_head io_buffers_cache;
	265
9ca9fb24	266	struct io_hash_table cancel_table_locked;
e27f928e	267	struct list_head cq_overflow_list;
9b797a37	268	struct io_alloc_cache apoll_cache;
43e0bbbd	269	struct io_alloc_cache netmsg_cache;
e27f928e JA	270	} ____cacheline_aligned_in_smp;
	271
	272	/* IRQ completion list, under ->completion_lock */
	273	struct io_wq_work_list locked_free_list;
	274	unsigned int locked_free_nr;
	275
	276	const struct cred sq_creds; / cred used for __io_sq_thread() */
	277	struct io_sq_data sq_data; / if using sq thread polling */
	278
	279	struct wait_queue_head sqo_sq_wait;
	280	struct list_head sqd_list;
	281
	282	unsigned long check_cq;
	283
6e73dffb PB	284	unsigned int file_alloc_start;
	285	unsigned int file_alloc_end;
	286
43e0bbbd JA	287	struct xarray personalities;
	288	u32 pers_next;
	289
e27f928e JA	290	struct {
	291	/*
	292	* We cache a range of free CQEs we can use, once exhausted it
	293	* should go through a slower range setup, see __io_get_cqe()
	294	*/
	295	struct io_uring_cqe *cqe_cached;
	296	struct io_uring_cqe *cqe_sentinel;
	297
	298	unsigned cached_cq_tail;
	299	unsigned cq_entries;
	300	struct io_ev_fd __rcu *io_ev_fd;
	301	struct wait_queue_head cq_wait;
	302	unsigned cq_extra;
e27f928e JA	303	} ____cacheline_aligned_in_smp;
	304
	305	struct {
	306	spinlock_t completion_lock;
	307
59b745bb	308	bool poll_multi_queue;
8751d154	309	atomic_t cq_wait_nr;
59b745bb	310
e27f928e JA	311	/*
	312	* ->iopoll_list is protected by the ctx->uring_lock for
	313	* io_uring instances that don't use IORING_SETUP_SQPOLL.
	314	* For SQPOLL, only the single threaded io_sq_thread() will
	315	* manipulate the list, hence no extra locking is needed there.
	316	*/
	317	struct io_wq_work_list iopoll_list;
e6f89be6	318	struct io_hash_table cancel_table;
e27f928e	319
c0e0d6ba DY	320	struct llist_head work_llist;
c0e0d6ba DY	321
e27f928e JA	322	struct list_head io_buffers_comp;
	323	} ____cacheline_aligned_in_smp;
	324
aff5b2df PB	325	/* timeouts */
	326	struct {
	327	spinlock_t timeout_lock;
	328	atomic_t cq_timeouts;
	329	struct list_head timeout_list;
	330	struct list_head ltimeout_list;
	331	unsigned cq_last_tm_flush;
	332	} ____cacheline_aligned_in_smp;
	333
e27f928e	334	/* Keep this last, we don't need it for the fast path */
7b235dd8	335	struct wait_queue_head poll_wq;
22eb2a3f PB	336	struct io_restriction restrictions;
	337
	338	/* slow path rsrc auxilary data, used by update/register */
22eb2a3f PB	339	struct io_mapped_ubuf *dummy_ubuf;
	340	struct io_rsrc_data *file_data;
	341	struct io_rsrc_data *buf_data;
	342
0a4813b1	343	/* protected by ->uring_lock */
22eb2a3f	344	struct list_head rsrc_ref_list;
9eae8655	345	struct io_alloc_cache rsrc_node_cache;
4ea15b56	346	struct wait_queue_head rsrc_quiesce_wq;
0b222eeb	347	unsigned rsrc_quiesce;
22eb2a3f PB	348
	349	struct list_head io_buffers_pages;
	350
	351	#if defined(CONFIG_UNIX)
	352	struct socket *ring_sock;
	353	#endif
	354	/* hashed buffered write serialization */
	355	struct io_wq_hash *hash_map;
	356
	357	/* Only used for accounting purposes */
	358	struct user_struct *user;
	359	struct mm_struct *mm_account;
	360
	361	/* ctx exit and cancelation */
	362	struct llist_head fallback_llist;
	363	struct delayed_work fallback_work;
	364	struct work_struct exit_work;
	365	struct list_head tctx_list;
	366	struct completion ref_comp;
	367
	368	/* io-wq management, e.g. thread count */
	369	u32 iowq_limits[2];
	370	bool iowq_limits_set;
	371
bca39f39	372	struct callback_head poll_wq_task_work;
22eb2a3f PB	373	struct list_head defer_list;
22eb2a3f PB	374	unsigned sq_thread_idle;
305bef98 PB	375	/* protected by ->completion_lock */
305bef98 PB	376	unsigned evfd_last_cq_tail;
e27f928e JA	377	};
e27f928e JA	378
a282967c PB	379	struct io_tw_state {
	380	/* ->uring_lock is taken, callbacks can use io_tw_lock to lock it */
	381	bool locked;
	382	};
	383
e27f928e JA	384	enum {
	385	REQ_F_FIXED_FILE_BIT = IOSQE_FIXED_FILE_BIT,
	386	REQ_F_IO_DRAIN_BIT = IOSQE_IO_DRAIN_BIT,
	387	REQ_F_LINK_BIT = IOSQE_IO_LINK_BIT,
	388	REQ_F_HARDLINK_BIT = IOSQE_IO_HARDLINK_BIT,
	389	REQ_F_FORCE_ASYNC_BIT = IOSQE_ASYNC_BIT,
	390	REQ_F_BUFFER_SELECT_BIT = IOSQE_BUFFER_SELECT_BIT,
	391	REQ_F_CQE_SKIP_BIT = IOSQE_CQE_SKIP_SUCCESS_BIT,
	392
	393	/* first byte is taken by user flags, shift it to not overlap */
	394	REQ_F_FAIL_BIT = 8,
	395	REQ_F_INFLIGHT_BIT,
	396	REQ_F_CUR_POS_BIT,
	397	REQ_F_NOWAIT_BIT,
	398	REQ_F_LINK_TIMEOUT_BIT,
	399	REQ_F_NEED_CLEANUP_BIT,
	400	REQ_F_POLLED_BIT,
	401	REQ_F_BUFFER_SELECTED_BIT,
	402	REQ_F_BUFFER_RING_BIT,
e27f928e JA	403	REQ_F_REISSUE_BIT,
	404	REQ_F_CREDS_BIT,
	405	REQ_F_REFCOUNT_BIT,
	406	REQ_F_ARM_LTIMEOUT_BIT,
	407	REQ_F_ASYNC_DATA_BIT,
	408	REQ_F_SKIP_LINK_CQES_BIT,
	409	REQ_F_SINGLE_POLL_BIT,
	410	REQ_F_DOUBLE_POLL_BIT,
	411	REQ_F_PARTIAL_IO_BIT,
e27f928e JA	412	REQ_F_APOLL_MULTISHOT_BIT,
e27f928e JA	413	REQ_F_CLEAR_POLLIN_BIT,
9ca9fb24	414	REQ_F_HASH_LOCKED_BIT,
e27f928e JA	415	/* keep async read/write and isreg together and in order */
	416	REQ_F_SUPPORT_NOWAIT_BIT,
	417	REQ_F_ISREG_BIT,
	418
	419	/* not a real bit, just to check we're not overflowing the space */
	420	__REQ_F_LAST_BIT,
	421	};
	422
	423	enum {
	424	/* ctx owns file */
	425	REQ_F_FIXED_FILE = BIT(REQ_F_FIXED_FILE_BIT),
	426	/* drain existing IO first */
	427	REQ_F_IO_DRAIN = BIT(REQ_F_IO_DRAIN_BIT),
	428	/* linked sqes */
	429	REQ_F_LINK = BIT(REQ_F_LINK_BIT),
	430	/* doesn't sever on completion < 0 */
	431	REQ_F_HARDLINK = BIT(REQ_F_HARDLINK_BIT),
	432	/* IOSQE_ASYNC */
	433	REQ_F_FORCE_ASYNC = BIT(REQ_F_FORCE_ASYNC_BIT),
	434	/* IOSQE_BUFFER_SELECT */
	435	REQ_F_BUFFER_SELECT = BIT(REQ_F_BUFFER_SELECT_BIT),
	436	/* IOSQE_CQE_SKIP_SUCCESS */
	437	REQ_F_CQE_SKIP = BIT(REQ_F_CQE_SKIP_BIT),
	438
	439	/* fail rest of links */
	440	REQ_F_FAIL = BIT(REQ_F_FAIL_BIT),
	441	/* on inflight list, should be cancelled and waited on exit reliably */
	442	REQ_F_INFLIGHT = BIT(REQ_F_INFLIGHT_BIT),
	443	/* read/write uses file position */
	444	REQ_F_CUR_POS = BIT(REQ_F_CUR_POS_BIT),
	445	/* must not punt to workers */
	446	REQ_F_NOWAIT = BIT(REQ_F_NOWAIT_BIT),
	447	/* has or had linked timeout */
	448	REQ_F_LINK_TIMEOUT = BIT(REQ_F_LINK_TIMEOUT_BIT),
	449	/* needs cleanup */
	450	REQ_F_NEED_CLEANUP = BIT(REQ_F_NEED_CLEANUP_BIT),
	451	/* already went through poll handler */
	452	REQ_F_POLLED = BIT(REQ_F_POLLED_BIT),
	453	/* buffer already selected */
	454	REQ_F_BUFFER_SELECTED = BIT(REQ_F_BUFFER_SELECTED_BIT),
	455	/* buffer selected from ring, needs commit */
	456	REQ_F_BUFFER_RING = BIT(REQ_F_BUFFER_RING_BIT),
e27f928e JA	457	/* caller should reissue async */
	458	REQ_F_REISSUE = BIT(REQ_F_REISSUE_BIT),
	459	/* supports async reads/writes */
	460	REQ_F_SUPPORT_NOWAIT = BIT(REQ_F_SUPPORT_NOWAIT_BIT),
	461	/* regular file */
	462	REQ_F_ISREG = BIT(REQ_F_ISREG_BIT),
	463	/* has creds assigned */
	464	REQ_F_CREDS = BIT(REQ_F_CREDS_BIT),
	465	/* skip refcounting if not set */
	466	REQ_F_REFCOUNT = BIT(REQ_F_REFCOUNT_BIT),
	467	/* there is a linked timeout that has to be armed */
	468	REQ_F_ARM_LTIMEOUT = BIT(REQ_F_ARM_LTIMEOUT_BIT),
	469	/* ->async_data allocated */
	470	REQ_F_ASYNC_DATA = BIT(REQ_F_ASYNC_DATA_BIT),
	471	/* don't post CQEs while failing linked requests */
	472	REQ_F_SKIP_LINK_CQES = BIT(REQ_F_SKIP_LINK_CQES_BIT),
	473	/* single poll may be active */
	474	REQ_F_SINGLE_POLL = BIT(REQ_F_SINGLE_POLL_BIT),
	475	/* double poll may active */
	476	REQ_F_DOUBLE_POLL = BIT(REQ_F_DOUBLE_POLL_BIT),
	477	/* request has already done partial IO */
	478	REQ_F_PARTIAL_IO = BIT(REQ_F_PARTIAL_IO_BIT),
	479	/* fast poll multishot mode */
	480	REQ_F_APOLL_MULTISHOT = BIT(REQ_F_APOLL_MULTISHOT_BIT),
e27f928e JA	481	/* recvmsg special flag, clear EPOLLIN */
e27f928e JA	482	REQ_F_CLEAR_POLLIN = BIT(REQ_F_CLEAR_POLLIN_BIT),
9ca9fb24 PB	483	/* hashed into ->cancel_hash_locked, protected by ->uring_lock */
9ca9fb24 PB	484	REQ_F_HASH_LOCKED = BIT(REQ_F_HASH_LOCKED_BIT),
e27f928e JA	485	};
e27f928e JA	486
a282967c	487	typedef void (io_req_tw_func_t)(struct io_kiocb req, struct io_tw_state *ts);
e27f928e JA	488
e27f928e JA	489	struct io_task_work {
3218e5d3	490	struct llist_node node;
e27f928e JA	491	io_req_tw_func_t func;
	492	};
	493
	494	struct io_cqe {
	495	__u64 user_data;
	496	__s32 res;
	497	/* fd initially, then cflags for completion */
	498	union {
	499	__u32 flags;
	500	int fd;
	501	};
	502	};
	503
	504	/*
	505	* Each request type overlays its private data structure on top of this one.
	506	* They must not exceed this one in size.
	507	*/
	508	struct io_cmd_data {
	509	struct file *file;
	510	/* each command gets 56 bytes of data */
	511	__u8 data[56];
	512	};
	513
f2ccb5ae SM	514	static inline void io_kiocb_cmd_sz_check(size_t cmd_sz)
	515	{
	516	BUILD_BUG_ON(cmd_sz > sizeof(struct io_cmd_data));
	517	}
	518	#define io_kiocb_to_cmd(req, cmd_type) ( \
	519	io_kiocb_cmd_sz_check(sizeof(cmd_type)) , \
	520	((cmd_type *)&(req)->cmd) \
	521	)
e27f928e JA	522	#define cmd_to_io_kiocb(ptr) ((struct io_kiocb *) ptr)
	523
	524	struct io_kiocb {
	525	union {
	526	/*
	527	* NOTE! Each of the io_kiocb union members has the file pointer
	528	* as the first entry in their struct definition. So you can
	529	* access the file pointer through any of the sub-structs,
	530	* or directly as just 'file' in this struct.
	531	*/
	532	struct file *file;
	533	struct io_cmd_data cmd;
	534	};
	535
	536	u8 opcode;
	537	/* polled IO has completed */
	538	u8 iopoll_completed;
	539	/*
	540	* Can be either a fixed buffer index, or used with provided buffers.
	541	* For the latter, before issue it points to the buffer group ID,
	542	* and after selection it points to the buffer ID itself.
	543	*/
	544	u16 buf_index;
	545	unsigned int flags;
	546
	547	struct io_cqe cqe;
	548
	549	struct io_ring_ctx *ctx;
	550	struct task_struct *task;
	551
	552	struct io_rsrc_node *rsrc_node;
	553
	554	union {
	555	/* store used ubuf, so we can prevent reloading */
	556	struct io_mapped_ubuf *imu;
	557
	558	/* stores selected buf, valid IFF REQ_F_BUFFER_SELECTED is set */
	559	struct io_buffer *kbuf;
	560
	561	/*
	562	* stores buffer ID for ring provided buffers, valid IFF
	563	* REQ_F_BUFFER_RING is set.
	564	*/
	565	struct io_buffer_list *buf_list;
	566	};
	567
	568	union {
	569	/* used by request caches, completion batching and iopoll */
	570	struct io_wq_work_node comp_list;
	571	/* cache ->apoll->events */
	572	__poll_t apoll_events;
	573	};
	574	atomic_t refs;
	575	atomic_t poll_refs;
	576	struct io_task_work io_task_work;
8751d154	577	unsigned nr_tw;
e27f928e	578	/* for polled requests, i.e. IORING_OP_POLL_ADD and async armed poll */
b24c5d75	579	struct hlist_node hash_node;
e27f928e JA	580	/* internal polling, see IORING_FEAT_FAST_POLL */
	581	struct async_poll *apoll;
	582	/* opcode allocated if it needs to store data for async defer */
	583	void *async_data;
	584	/* linked requests, IFF REQ_F_HARDLINK or REQ_F_LINK are set */
	585	struct io_kiocb *link;
	586	/* custom credentials, valid IFF REQ_F_CREDS is set */
	587	const struct cred *creds;
	588	struct io_wq_work work;
b24c5d75 PB	589
	590	struct {
	591	u64 extra1;
	592	u64 extra2;
	593	} big_cqe;
e27f928e JA	594	};
e27f928e JA	595
a4ad4f74 JA	596	struct io_overflow_cqe {
	597	struct list_head list;
	598	struct io_uring_cqe cqe;
	599	};
	600
e27f928e	601	#endif