liburing API

/*
* Library interface to io_uring
*/
struct io_uring_sq {
   unsigned *khead;
   unsigned *ktail;
   unsigned *kring_mask;
   unsigned *kring_entries;
   unsigned *kflags;
   unsigned *kdropped;
   unsigned *array;
   struct io_uring_sqe *sqes;

   unsigned sqe_head;
   unsigned sqe_tail;

   size_t ring_sz;
   void *ring_ptr;

   unsigned pad[4];
};

struct io_uring_cq {
   unsigned *khead;
   unsigned *ktail;
   unsigned *kring_mask;
   unsigned *kring_entries;
   unsigned *kflags;
   unsigned *koverflow;
   struct io_uring_cqe *cqes;

   size_t ring_sz;
   void *ring_ptr;

   unsigned pad[4];
};

struct io_uring {
   struct io_uring_sq sq;
   struct io_uring_cq cq;
   unsigned flags;
   int ring_fd;

   unsigned features;
   unsigned pad[3];
};



/*
* Library interface
*/

extern struct io_uring_probe *io_uring_get_probe_ring(struct io_uring *ring);
extern struct io_uring_probe *io_uring_get_probe(void);
extern void io_uring_free_probe(struct io_uring_probe *probe);
static inline int io_uring_opcode_supported(struct io_uring_probe *p, int op);

extern int io_uring_queue_init_params(unsigned entries, struct io_uring *ring, struct io_uring_params *p);

extern int io_uring_queue_init(unsigned entries, struct io_uring *ring, unsigned flags);    //初始化struct io_uring，函数内部会调用io_uring_queue_init_params。io_uring 是一个循环队列（ring_buffer）。第一个参数 entries 表示队列大小（实际空间可能比用户指定的大）；第二个参数 ring 就是需要初始化的 io_uring 结构指针；第三个参数 flags 是标志参数，无特殊需要传 0 即可。

extern int io_uring_queue_mmap(int fd, struct io_uring_params *p, struct io_uring *ring);
extern int io_uring_ring_dontfork(struct io_uring *ring);
extern void io_uring_queue_exit(struct io_uring *ring);
unsigned io_uring_peek_batch_cqe(struct io_uring *ring, struct io_uring_cqe **cqes, unsigned count);
extern int io_uring_wait_cqes(struct io_uring *ring, struct io_uring_cqe **cqe_ptr, unsigned wait_nr, struct __kernel_timespec *ts, sigset_t *sigmask);
extern int io_uring_wait_cqe_timeout(struct io_uring *ring, struct io_uring_cqe **cqe_ptr, struct __kernel_timespec *ts);
extern int io_uring_submit(struct io_uring *ring);
extern int io_uring_submit_and_wait(struct io_uring *ring, unsigned wait_nr);
extern struct io_uring_sqe *io_uring_get_sqe(struct io_uring *ring);

extern int io_uring_register_buffers(struct io_uring *ring, const struct iovec *iovecs, unsigned nr_iovecs);
extern int io_uring_unregister_buffers(struct io_uring *ring);
extern int io_uring_register_files(struct io_uring *ring, const int *files, unsigned nr_files);
extern int io_uring_unregister_files(struct io_uring *ring);
extern int io_uring_register_files_update(struct io_uring *ring, unsigned off, int *files, unsigned nr_files);
extern int io_uring_register_eventfd(struct io_uring *ring, int fd);
extern int io_uring_register_eventfd_async(struct io_uring *ring, int fd);
extern int io_uring_unregister_eventfd(struct io_uring *ring);
extern int io_uring_register_probe(struct io_uring *ring, struct io_uring_probe *p, unsigned nr);
extern int io_uring_register_personality(struct io_uring *ring);
extern int io_uring_unregister_personality(struct io_uring *ring, int id);
extern int io_uring_register_restrictions(struct io_uring *ring, struct io_uring_restriction *res, unsigned int nr_res);
extern int io_uring_enable_rings(struct io_uring *ring);
extern int __io_uring_sqring_wait(struct io_uring *ring);

extern int __io_uring_get_cqe(struct io_uring *ring, struct io_uring_cqe **cqe_ptr, unsigned submit, unsigned wait_nr, sigset_t *sigmask);
#define io_uring_for_each_cqe(ring, head, cqe)

static inline void io_uring_cq_advance(struct io_uring *ring, unsigned nr)
static inline void io_uring_cqe_seen(struct io_uring *ring, struct io_uring_cqe *cqe)
static inline void io_uring_sqe_set_data(struct io_uring_sqe *sqe, void *data)
static inline void *io_uring_cqe_get_data(const struct io_uring_cqe *cqe)
static inline void io_uring_sqe_set_flags(struct io_uring_sqe *sqe,
static inline void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd, const void *addr, unsigned len, __u64 offset)
static inline void io_uring_prep_splice(struct io_uring_sqe *sqe, int fd_in, int64_t off_in, int fd_out, int64_t off_out, unsigned int nbytes, unsigned int splice_flags)
static inline void io_uring_prep_tee(struct io_uring_sqe *sqe, int fd_in, int fd_out, unsigned int nbytes, unsigned int splice_flags)

io_uring使用

1、首先需要创建io_uring struct
2、然后使用io_uring_queue_init(unsigned entries, struct io_uring *ring, unsigned flags);初始化io_uring struct，io_uring_queue_init函数内部会调用函数内部会调用io_uring_queue_init_params。

io_uring struct在liburing中定义如下
struct io_uring {
   struct io_uring_sq sq;
   struct io_uring_cq cq;
   unsigned flags;
   int ring_fd;
   unsigned features;
   unsigned pad[3];
};

3、获取并sqe struct
一个 sqe（submission queue entry）代表一次 IO 请求，占用循环队列一个空位。io_uring 队列满时 io_uring_get_sqe 会返回 NULL，注意错误处理。注意这里的队列是指未提交的请求，已提交的（但未完成）请求不占位置。
struct io_uring_sqe *sqe = io_uring_get_sqe(&ring);

4、初始化sqe struct
io_uring_prep_readv 或 io_uring_prep_writev 初始化 sqe struct。

io_uring_prep_readv(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, off_t offset)会调用io_uring_prep_rw(IORING_OP_READV, sqe, fd, iovecs, nr_vecs, offset)
io_uring_prep_rw的定义
static inline void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd,
                const void *addr, unsigned len,
                __u64 offset)
{
   sqe->opcode = op;
   sqe->flags = 0;
   sqe->ioprio = 0;
   sqe->fd = fd;
   sqe->off = offset;
   sqe->addr = (unsigned long) addr;
   sqe->len = len;
   sqe->rw_flags = 0;
   sqe->user_data = 0;
   sqe->__pad2[0] = sqe->__pad2[1] = sqe->__pad2[2] = 0;
}

io_uring_prep_writev(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, off_t offset)会调用io_uring_prep_rw(IORING_OP_WRITEV, sqe, fd, iovecs, nr_vecs, offset)
io_uring_prep_rw的定义
static inline void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd,
                const void *addr, unsigned len,
                __u64 offset)
{
   sqe->opcode = op;
   sqe->flags = 0;
   sqe->ioprio = 0;
   sqe->fd = fd;
   sqe->off = offset;
   sqe->addr = (unsigned long) addr;
   sqe->len = len;
   sqe->rw_flags = 0;
   sqe->user_data = 0;
   sqe->__pad2[0] = sqe->__pad2[1] = sqe->__pad2[2] = 0;
}

sqe struct的定义
/*
* 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;
      __u32     rename_flags;
      __u32     unlink_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];
   };
};

5、提交sqe（也就是提交IO）
/*
* Submit sqes acquired from io_uring_get_sqe() to the kernel.
*
* Returns number of sqes submitted
*/
int io_uring_submit(struct io_uring *ring)
{
   return __io_uring_submit_and_wait(ring, 0);
}

static int __io_uring_submit_and_wait(struct io_uring *ring, unsigned wait_nr)
{
   return __io_uring_submit(ring, __io_uring_flush_sq(ring), wait_nr);
}

/*
* Submit sqes acquired from io_uring_get_sqe() to the kernel.
*
* Returns number of sqes submitted
*/
static int __io_uring_submit(struct io_uring *ring, unsigned submitted,
              unsigned wait_nr)
{
   unsigned flags;
   int ret;


   flags = 0;
   if (sq_ring_needs_enter(ring, &flags) || wait_nr) {
      if (wait_nr || (ring->flags & IORING_SETUP_IOPOLL))
         flags |= IORING_ENTER_GETEVENTS;


      ret = __sys_io_uring_enter(ring->ring_fd, submitted, wait_nr,
                  flags, NULL);
      if (ret < 0)
         return -errno;
   } else
      ret = submitted;


   return ret;
}

int __sys_io_uring_enter(int fd, unsigned to_submit, unsigned min_complete,
          unsigned flags, sigset_t *sig)
{
   return __sys_io_uring_enter2(fd, to_submit, min_complete, flags, sig,
               _NSIG / 8);
}

int __sys_io_uring_enter2(int fd, unsigned to_submit, unsigned min_complete,
          unsigned flags, sigset_t *sig, int sz)
{
   return syscall(__NR_io_uring_enter, fd, to_submit, min_complete,
         flags, sig, sz);
}

6、提交sqe后，我们只需要等待io完成

鸣谢：
https://segmentfault.com/a/1190000019300089

demo：
fio中的代码 https://github.com/lnsyyj/Codes/blob/master/iouring/io_uring-test.c