[fio.git] / os / linux / io_uring.h

/* SPDX-License-Identifier: (GPL-2.0 WITH Linux-syscall-note) OR MIT */
/*
 * Header file for the io_uring interface.
 *
 * Copyright (C) 2019 Jens Axboe
 * Copyright (C) 2019 Christoph Hellwig
 */
#ifndef LINUX_IO_URING_H
#define LINUX_IO_URING_H

#include <linux/fs.h>
#include <linux/types.h>

#ifdef __cplusplus
extern "C" {
#endif

/*
 * IO submission data structure (Submission Queue Entry)
 */
struct io_uring_sqe {
	__u8	opcode;		/* type of operation for this sqe */
	__u8	flags;		/* IOSQE_ flags */
	__u16	ioprio;		/* ioprio for the request */
	__s32	fd;		/* file descriptor to do IO on */
	union {
		__u64	off;	/* offset into file */
		__u64	addr2;
	};
	union {
		__u64	addr;	/* pointer to buffer or iovecs */
		__u64	splice_off_in;
	};
	__u32	len;		/* buffer size or number of iovecs */
	union {
		__kernel_rwf_t	rw_flags;
		__u32		fsync_flags;
		__u16		poll_events;	/* compatibility */
		__u32		poll32_events;	/* word-reversed for BE */
		__u32		sync_range_flags;
		__u32		msg_flags;
		__u32		timeout_flags;
		__u32		accept_flags;
		__u32		cancel_flags;
		__u32		open_flags;
		__u32		statx_flags;
		__u32		fadvise_advice;
		__u32		splice_flags;
	};
	__u64	user_data;	/* data to be passed back at completion time */
	union {
		struct {
			/* pack this to avoid bogus arm OABI complaints */
			union {
				/* index into fixed buffers, if used */
				__u16	buf_index;
				/* for grouped buffer selection */
				__u16	buf_group;
			} __attribute__((packed));
			/* personality to use, if used */
			__u16	personality;
			__s32	splice_fd_in;
		};
		__u64	__pad2[3];
	};
};

enum {
	IOSQE_FIXED_FILE_BIT,
	IOSQE_IO_DRAIN_BIT,
	IOSQE_IO_LINK_BIT,
	IOSQE_IO_HARDLINK_BIT,
	IOSQE_ASYNC_BIT,
	IOSQE_BUFFER_SELECT_BIT,
};

/*
 * sqe->flags
 */
/* use fixed fileset */
#define IOSQE_FIXED_FILE	(1U << IOSQE_FIXED_FILE_BIT)
/* issue after inflight IO */
#define IOSQE_IO_DRAIN		(1U << IOSQE_IO_DRAIN_BIT)
/* links next sqe */
#define IOSQE_IO_LINK		(1U << IOSQE_IO_LINK_BIT)
/* like LINK, but stronger */
#define IOSQE_IO_HARDLINK	(1U << IOSQE_IO_HARDLINK_BIT)
/* always go async */
#define IOSQE_ASYNC		(1U << IOSQE_ASYNC_BIT)
/* select buffer from sqe->buf_group */
#define IOSQE_BUFFER_SELECT	(1U << IOSQE_BUFFER_SELECT_BIT)

/*
 * io_uring_setup() flags
 */
#define IORING_SETUP_IOPOLL	(1U << 0)	/* io_context is polled */
#define IORING_SETUP_SQPOLL	(1U << 1)	/* SQ poll thread */
#define IORING_SETUP_SQ_AFF	(1U << 2)	/* sq_thread_cpu is valid */
#define IORING_SETUP_CQSIZE	(1U << 3)	/* app defines CQ size */
#define IORING_SETUP_CLAMP	(1U << 4)	/* clamp SQ/CQ ring sizes */
#define IORING_SETUP_ATTACH_WQ	(1U << 5)	/* attach to existing wq */

enum {
	IORING_OP_NOP,
	IORING_OP_READV,
	IORING_OP_WRITEV,
	IORING_OP_FSYNC,
	IORING_OP_READ_FIXED,
	IORING_OP_WRITE_FIXED,
	IORING_OP_POLL_ADD,
	IORING_OP_POLL_REMOVE,
	IORING_OP_SYNC_FILE_RANGE,
	IORING_OP_SENDMSG,
	IORING_OP_RECVMSG,
	IORING_OP_TIMEOUT,
	IORING_OP_TIMEOUT_REMOVE,
	IORING_OP_ACCEPT,
	IORING_OP_ASYNC_CANCEL,
	IORING_OP_LINK_TIMEOUT,
	IORING_OP_CONNECT,
	IORING_OP_FALLOCATE,
	IORING_OP_OPENAT,
	IORING_OP_CLOSE,
	IORING_OP_FILES_UPDATE,
	IORING_OP_STATX,
	IORING_OP_READ,
	IORING_OP_WRITE,
	IORING_OP_FADVISE,
	IORING_OP_MADVISE,
	IORING_OP_SEND,
	IORING_OP_RECV,
	IORING_OP_OPENAT2,
	IORING_OP_EPOLL_CTL,
	IORING_OP_SPLICE,
	IORING_OP_PROVIDE_BUFFERS,
	IORING_OP_REMOVE_BUFFERS,
	IORING_OP_TEE,

	/* this goes last, obviously */
	IORING_OP_LAST,
};

/*
 * sqe->fsync_flags
 */
#define IORING_FSYNC_DATASYNC	(1U << 0)

/*
 * sqe->timeout_flags
 */
#define IORING_TIMEOUT_ABS	(1U << 0)

/*
 * sqe->splice_flags
 * extends splice(2) flags
 */
#define SPLICE_F_FD_IN_FIXED	(1U << 31) /* the last bit of __u32 */

/*
 * IO completion data structure (Completion Queue Entry)
 */
struct io_uring_cqe {
	__u64	user_data;	/* sqe->data submission passed back */
	__s32	res;		/* result code for this event */
	__u32	flags;
};

/*
 * cqe->flags
 *
 * IORING_CQE_F_BUFFER	If set, the upper 16 bits are the buffer ID
 */
#define IORING_CQE_F_BUFFER		(1U << 0)

enum {
	IORING_CQE_BUFFER_SHIFT		= 16,
};

/*
 * Magic offsets for the application to mmap the data it needs
 */
#define IORING_OFF_SQ_RING		0ULL
#define IORING_OFF_CQ_RING		0x8000000ULL
#define IORING_OFF_SQES			0x10000000ULL

/*
 * Filled with the offset for mmap(2)
 */
struct io_sqring_offsets {
	__u32 head;
	__u32 tail;
	__u32 ring_mask;
	__u32 ring_entries;
	__u32 flags;
	__u32 dropped;
	__u32 array;
	__u32 resv1;
	__u64 resv2;
};

/*
 * sq_ring->flags
 */
#define IORING_SQ_NEED_WAKEUP	(1U << 0) /* needs io_uring_enter wakeup */
#define IORING_SQ_CQ_OVERFLOW	(1U << 1) /* CQ ring is overflown */

struct io_cqring_offsets {
	__u32 head;
	__u32 tail;
	__u32 ring_mask;
	__u32 ring_entries;
	__u32 overflow;
	__u32 cqes;
	__u32 flags;
	__u32 resv1;
	__u64 resv2;
};

/*
 * cq_ring->flags
 */

/* disable eventfd notifications */
#define IORING_CQ_EVENTFD_DISABLED	(1U << 0)

/*
 * io_uring_enter(2) flags
 */
#define IORING_ENTER_GETEVENTS	(1U << 0)
#define IORING_ENTER_SQ_WAKEUP	(1U << 1)

/*
 * Passed in for io_uring_setup(2). Copied back with updated info on success
 */
struct io_uring_params {
	__u32 sq_entries;
	__u32 cq_entries;
	__u32 flags;
	__u32 sq_thread_cpu;
	__u32 sq_thread_idle;
	__u32 features;
	__u32 wq_fd;
	__u32 resv[3];
	struct io_sqring_offsets sq_off;
	struct io_cqring_offsets cq_off;
};

/*
 * io_uring_params->features flags
 */
#define IORING_FEAT_SINGLE_MMAP		(1U << 0)
#define IORING_FEAT_NODROP		(1U << 1)
#define IORING_FEAT_SUBMIT_STABLE	(1U << 2)
#define IORING_FEAT_RW_CUR_POS		(1U << 3)
#define IORING_FEAT_CUR_PERSONALITY	(1U << 4)
#define IORING_FEAT_FAST_POLL		(1U << 5)
#define IORING_FEAT_POLL_32BITS 	(1U << 6)

/*
 * io_uring_register(2) opcodes and arguments
 */
#define IORING_REGISTER_BUFFERS		0
#define IORING_UNREGISTER_BUFFERS	1
#define IORING_REGISTER_FILES		2
#define IORING_UNREGISTER_FILES		3
#define IORING_REGISTER_EVENTFD		4
#define IORING_UNREGISTER_EVENTFD	5
#define IORING_REGISTER_FILES_UPDATE	6
#define IORING_REGISTER_EVENTFD_ASYNC	7
#define IORING_REGISTER_PROBE		8
#define IORING_REGISTER_PERSONALITY	9
#define IORING_UNREGISTER_PERSONALITY	10

struct io_uring_files_update {
	__u32 offset;
	__u32 resv;
	__aligned_u64 /* __s32 * */ fds;
};

#define IO_URING_OP_SUPPORTED	(1U << 0)

struct io_uring_probe_op {
	__u8 op;
	__u8 resv;
	__u16 flags;	/* IO_URING_OP_* flags */
	__u32 resv2;
};

struct io_uring_probe {
	__u8 last_op;	/* last opcode supported */
	__u8 ops_len;	/* length of ops[] array below */
	__u16 resv;
	__u32 resv2[3];
	struct io_uring_probe_op ops[0];
};

#ifdef __cplusplus
}
#endif

#endif
Commit	Line	Data
556d8415	1	/* SPDX-License-Identifier: (GPL-2.0 WITH Linux-syscall-note) OR MIT */
e2239016 JA	2	/*
	3	* Header file for the io_uring interface.
	4	*
	5	* Copyright (C) 2019 Jens Axboe
	6	* Copyright (C) 2019 Christoph Hellwig
	7	*/
	8	#ifndef LINUX_IO_URING_H
	9	#define LINUX_IO_URING_H
e31b8288	10
6e70fd30	11	#include <linux/fs.h>
e2239016	12	#include <linux/types.h>
6e70fd30	13
556d8415 JA	14	#ifdef __cplusplus
	15	extern "C" {
	16	#endif
	17
e31b8288	18	/*
f0403f94	19	* IO submission data structure (Submission Queue Entry)
e31b8288	20	*/
f0403f94	21	struct io_uring_sqe {
a7086591	22	__u8 opcode; /* type of operation for this sqe */
48e698fa	23	__u8 flags; /* IOSQE_ flags */
a7086591 JA	24	__u16 ioprio; /* ioprio for the request */
a7086591 JA	25	__s32 fd; /* file descriptor to do IO on */
b10b1e70 JA	26	union {
	27	__u64 off; /* offset into file */
	28	__u64 addr2;
	29	};
556d8415 JA	30	union {
	31	__u64 addr; /* pointer to buffer or iovecs */
	32	__u64 splice_off_in;
	33	};
a7086591	34	__u32 len; /* buffer size or number of iovecs */
e31b8288 JA	35	union {
e31b8288 JA	36	__kernel_rwf_t rw_flags;
48e698fa	37	__u32 fsync_flags;
556d8415 JA	38	__u16 poll_events; /* compatibility */
556d8415 JA	39	__u32 poll32_events; /* word-reversed for BE */
a819dfb6	40	__u32 sync_range_flags;
b10b1e70 JA	41	__u32 msg_flags;
	42	__u32 timeout_flags;
	43	__u32 accept_flags;
	44	__u32 cancel_flags;
	45	__u32 open_flags;
	46	__u32 statx_flags;
556d8415 JA	47	__u32 fadvise_advice;
556d8415 JA	48	__u32 splice_flags;
e31b8288	49	};
48e698fa	50	__u64 user_data; /* data to be passed back at completion time */
f3e769a4	51	union {
556d8415 JA	52	struct {
	53	/* pack this to avoid bogus arm OABI complaints */
	54	union {
	55	/* index into fixed buffers, if used */
	56	__u16 buf_index;
	57	/* for grouped buffer selection */
	58	__u16 buf_group;
	59	} __attribute__((packed));
	60	/* personality to use, if used */
	61	__u16 personality;
	62	__s32 splice_fd_in;
	63	};
f3e769a4 JA	64	__u64 __pad2[3];
f3e769a4 JA	65	};
e31b8288 JA	66	};
e31b8288 JA	67
556d8415 JA	68	enum {
	69	IOSQE_FIXED_FILE_BIT,
	70	IOSQE_IO_DRAIN_BIT,
	71	IOSQE_IO_LINK_BIT,
	72	IOSQE_IO_HARDLINK_BIT,
	73	IOSQE_ASYNC_BIT,
	74	IOSQE_BUFFER_SELECT_BIT,
	75	};
	76
2ea53ca3 JA	77	/*
	78	* sqe->flags
	79	*/
556d8415 JA	80	/* use fixed fileset */
	81	#define IOSQE_FIXED_FILE (1U << IOSQE_FIXED_FILE_BIT)
	82	/* issue after inflight IO */
	83	#define IOSQE_IO_DRAIN (1U << IOSQE_IO_DRAIN_BIT)
	84	/* links next sqe */
	85	#define IOSQE_IO_LINK (1U << IOSQE_IO_LINK_BIT)
	86	/* like LINK, but stronger */
	87	#define IOSQE_IO_HARDLINK (1U << IOSQE_IO_HARDLINK_BIT)
	88	/* always go async */
	89	#define IOSQE_ASYNC (1U << IOSQE_ASYNC_BIT)
	90	/* select buffer from sqe->buf_group */
	91	#define IOSQE_BUFFER_SELECT (1U << IOSQE_BUFFER_SELECT_BIT)
2ea53ca3	92
e31b8288 JA	93	/*
	94	* io_uring_setup() flags
	95	*/
e9d4aa07 JA	96	#define IORING_SETUP_IOPOLL (1U << 0) /* io_context is polled */
	97	#define IORING_SETUP_SQPOLL (1U << 1) /* SQ poll thread */
	98	#define IORING_SETUP_SQ_AFF (1U << 2) /* sq_thread_cpu is valid */
b10b1e70	99	#define IORING_SETUP_CQSIZE (1U << 3) /* app defines CQ size */
556d8415 JA	100	#define IORING_SETUP_CLAMP (1U << 4) /* clamp SQ/CQ ring sizes */
556d8415 JA	101	#define IORING_SETUP_ATTACH_WQ (1U << 5) /* attach to existing wq */
b10b1e70 JA	102
	103	enum {
	104	IORING_OP_NOP,
	105	IORING_OP_READV,
	106	IORING_OP_WRITEV,
	107	IORING_OP_FSYNC,
	108	IORING_OP_READ_FIXED,
	109	IORING_OP_WRITE_FIXED,
	110	IORING_OP_POLL_ADD,
	111	IORING_OP_POLL_REMOVE,
	112	IORING_OP_SYNC_FILE_RANGE,
	113	IORING_OP_SENDMSG,
	114	IORING_OP_RECVMSG,
	115	IORING_OP_TIMEOUT,
	116	IORING_OP_TIMEOUT_REMOVE,
	117	IORING_OP_ACCEPT,
	118	IORING_OP_ASYNC_CANCEL,
	119	IORING_OP_LINK_TIMEOUT,
	120	IORING_OP_CONNECT,
	121	IORING_OP_FALLOCATE,
	122	IORING_OP_OPENAT,
	123	IORING_OP_CLOSE,
	124	IORING_OP_FILES_UPDATE,
	125	IORING_OP_STATX,
	126	IORING_OP_READ,
	127	IORING_OP_WRITE,
556d8415 JA	128	IORING_OP_FADVISE,
	129	IORING_OP_MADVISE,
	130	IORING_OP_SEND,
	131	IORING_OP_RECV,
	132	IORING_OP_OPENAT2,
	133	IORING_OP_EPOLL_CTL,
	134	IORING_OP_SPLICE,
	135	IORING_OP_PROVIDE_BUFFERS,
	136	IORING_OP_REMOVE_BUFFERS,
	137	IORING_OP_TEE,
e31b8288	138
b10b1e70 JA	139	/* this goes last, obviously */
	140	IORING_OP_LAST,
	141	};
48e698fa JA	142
	143	/*
	144	* sqe->fsync_flags
	145	*/
e9d4aa07	146	#define IORING_FSYNC_DATASYNC (1U << 0)
e31b8288	147
b10b1e70 JA	148	/*
	149	* sqe->timeout_flags
	150	*/
	151	#define IORING_TIMEOUT_ABS (1U << 0)
	152
556d8415 JA	153	/*
	154	* sqe->splice_flags
	155	* extends splice(2) flags
	156	*/
	157	#define SPLICE_F_FD_IN_FIXED (1U << 31) /* the last bit of __u32 */
	158
e31b8288	159	/*
f0403f94	160	* IO completion data structure (Completion Queue Entry)
e31b8288	161	*/
f0403f94	162	struct io_uring_cqe {
48e698fa	163	__u64 user_data; /* sqe->data submission passed back */
e2239016 JA	164	__s32 res; /* result code for this event */
e2239016 JA	165	__u32 flags;
e31b8288 JA	166	};
e31b8288 JA	167
556d8415 JA	168	/*
	169	* cqe->flags
	170	*
	171	* IORING_CQE_F_BUFFER If set, the upper 16 bits are the buffer ID
	172	*/
	173	#define IORING_CQE_F_BUFFER (1U << 0)
	174
	175	enum {
	176	IORING_CQE_BUFFER_SHIFT = 16,
	177	};
	178
e31b8288 JA	179	/*
	180	* Magic offsets for the application to mmap the data it needs
	181	*/
	182	#define IORING_OFF_SQ_RING 0ULL
	183	#define IORING_OFF_CQ_RING 0x8000000ULL
f0403f94	184	#define IORING_OFF_SQES 0x10000000ULL
e31b8288 JA	185
	186	/*
	187	* Filled with the offset for mmap(2)
	188	*/
	189	struct io_sqring_offsets {
e2239016 JA	190	__u32 head;
	191	__u32 tail;
	192	__u32 ring_mask;
	193	__u32 ring_entries;
	194	__u32 flags;
	195	__u32 dropped;
	196	__u32 array;
66169493 JA	197	__u32 resv1;
66169493 JA	198	__u64 resv2;
e31b8288 JA	199	};
e31b8288 JA	200
48e698fa JA	201	/*
	202	* sq_ring->flags
	203	*/
e9d4aa07	204	#define IORING_SQ_NEED_WAKEUP (1U << 0) /* needs io_uring_enter wakeup */
556d8415	205	#define IORING_SQ_CQ_OVERFLOW (1U << 1) /* CQ ring is overflown */
e31b8288 JA	206
e31b8288 JA	207	struct io_cqring_offsets {
e2239016 JA	208	__u32 head;
	209	__u32 tail;
	210	__u32 ring_mask;
	211	__u32 ring_entries;
	212	__u32 overflow;
f0403f94	213	__u32 cqes;
556d8415 JA	214	__u32 flags;
	215	__u32 resv1;
	216	__u64 resv2;
e31b8288 JA	217	};
e31b8288 JA	218
556d8415 JA	219	/*
	220	* cq_ring->flags
	221	*/
	222
	223	/* disable eventfd notifications */
	224	#define IORING_CQ_EVENTFD_DISABLED (1U << 0)
	225
e2239016 JA	226	/*
	227	* io_uring_enter(2) flags
	228	*/
e9d4aa07 JA	229	#define IORING_ENTER_GETEVENTS (1U << 0)
e9d4aa07 JA	230	#define IORING_ENTER_SQ_WAKEUP (1U << 1)
e31b8288 JA	231
	232	/*
	233	* Passed in for io_uring_setup(2). Copied back with updated info on success
	234	*/
	235	struct io_uring_params {
e2239016 JA	236	__u32 sq_entries;
	237	__u32 cq_entries;
	238	__u32 flags;
e9d4aa07 JA	239	__u32 sq_thread_cpu;
e9d4aa07 JA	240	__u32 sq_thread_idle;
b10b1e70	241	__u32 features;
556d8415 JA	242	__u32 wq_fd;
556d8415 JA	243	__u32 resv[3];
e31b8288 JA	244	struct io_sqring_offsets sq_off;
	245	struct io_cqring_offsets cq_off;
	246	};
	247
b10b1e70 JA	248	/*
	249	* io_uring_params->features flags
	250	*/
	251	#define IORING_FEAT_SINGLE_MMAP (1U << 0)
	252	#define IORING_FEAT_NODROP (1U << 1)
	253	#define IORING_FEAT_SUBMIT_STABLE (1U << 2)
556d8415 JA	254	#define IORING_FEAT_RW_CUR_POS (1U << 3)
	255	#define IORING_FEAT_CUR_PERSONALITY (1U << 4)
	256	#define IORING_FEAT_FAST_POLL (1U << 5)
	257	#define IORING_FEAT_POLL_32BITS (1U << 6)
b10b1e70	258
2ea53ca3 JA	259	/*
	260	* io_uring_register(2) opcodes and arguments
	261	*/
	262	#define IORING_REGISTER_BUFFERS 0
	263	#define IORING_UNREGISTER_BUFFERS 1
a7abc9fb JA	264	#define IORING_REGISTER_FILES 2
a7abc9fb JA	265	#define IORING_UNREGISTER_FILES 3
a819dfb6 JA	266	#define IORING_REGISTER_EVENTFD 4
a819dfb6 JA	267	#define IORING_UNREGISTER_EVENTFD 5
b10b1e70	268	#define IORING_REGISTER_FILES_UPDATE 6
556d8415 JA	269	#define IORING_REGISTER_EVENTFD_ASYNC 7
	270	#define IORING_REGISTER_PROBE 8
	271	#define IORING_REGISTER_PERSONALITY 9
	272	#define IORING_UNREGISTER_PERSONALITY 10
b10b1e70 JA	273
	274	struct io_uring_files_update {
	275	__u32 offset;
556d8415 JA	276	__u32 resv;
556d8415 JA	277	__aligned_u64 /* __s32 * */ fds;
b10b1e70	278	};
2ea53ca3	279
556d8415 JA	280	#define IO_URING_OP_SUPPORTED (1U << 0)
	281
	282	struct io_uring_probe_op {
	283	__u8 op;
	284	__u8 resv;
	285	__u16 flags; /* IO_URING_OP_* flags */
	286	__u32 resv2;
	287	};
	288
	289	struct io_uring_probe {
	290	__u8 last_op; /* last opcode supported */
	291	__u8 ops_len; /* length of ops[] array below */
	292	__u16 resv;
	293	__u32 resv2[3];
	294	struct io_uring_probe_op ops[0];
	295	};
	296
	297	#ifdef __cplusplus
	298	}
	299	#endif
	300
e31b8288	301	#endif