文档

API 使用

https://nick-black.com/dankwiki/index.php/Io_uring

这里里面的东西全部都需要背下来吧

io uring 一共支持哪些系统调用

一共就是三个:

int io_uring_setup(u32 entries, struct io_uring_params *p);
int io_uring_register(unsigned fd, unsigned opcode, void *arg, unsigned int nr_args);
int io_uring_enter(unsigned fd, u32 to_submit, u32 min_complete, u32 flags, const void* argp, size_t argsz);

所以，当运行起来的时候，只会去调用 io_uring_enter 的

liburing 提供的两个接口都是回去调用 io_uring_enter :

io_uring_enter 参数 flags 的含义

https://manpages.debian.org/unstable/liburing-dev/io_uring_enter.2.en.html

/*
 * io_uring_enter(2) flags
 */
#define IORING_ENTER_GETEVENTS		(1U << 0)
#define IORING_ENTER_SQ_WAKEUP		(1U << 1)
#define IORING_ENTER_SQ_WAIT		(1U << 2)
#define IORING_ENTER_EXT_ARG		(1U << 3)
#define IORING_ENTER_REGISTERED_RING	(1U << 4)
#define IORING_ENTER_ABS_TIMER		(1U << 5)
#define IORING_ENTER_EXT_ARG_REG	(1U << 6)
#define IORING_ENTER_NO_IOWAIT		(1U << 7)

io_uring_setup(2)

https://man7.org/linux/man-pages/man2/io_uring_setup.2.html

io_uring_register(2) 支持的功能

主要是这几类:

  1. Ring : IORING_REGISTER_RING_FDS (不能理解!)
  2. Buffers : 固定 buffer I/O（高性能）
  3. File descriptors : 内核无需每次通过 fdtable 查找 file 结构，提升性能。
    • 但是通过 fd 查询找到文件的过程比较很低效的

https://man7.org/linux/man-pages/man2/io_uring_register.2.html 完整结果如下:

/*
 * io_uring_register(2) opcodes and arguments
 */
enum io_uring_register_op {
	IORING_REGISTER_BUFFERS			= 0,
	IORING_UNREGISTER_BUFFERS		= 1,
	IORING_REGISTER_FILES			= 2,
	IORING_UNREGISTER_FILES			= 3,
	IORING_REGISTER_EVENTFD			= 4,
	IORING_UNREGISTER_EVENTFD		= 5,
	IORING_REGISTER_FILES_UPDATE		= 6,
	IORING_REGISTER_EVENTFD_ASYNC		= 7,
	IORING_REGISTER_PROBE			= 8,
	IORING_REGISTER_PERSONALITY		= 9,
	IORING_UNREGISTER_PERSONALITY		= 10,
	IORING_REGISTER_RESTRICTIONS		= 11,
	IORING_REGISTER_ENABLE_RINGS		= 12,

	/* extended with tagging */
	IORING_REGISTER_FILES2			= 13,
	IORING_REGISTER_FILES_UPDATE2		= 14,
	IORING_REGISTER_BUFFERS2		= 15,
	IORING_REGISTER_BUFFERS_UPDATE		= 16,

	/* set/clear io-wq thread affinities */
	IORING_REGISTER_IOWQ_AFF		= 17,
	IORING_UNREGISTER_IOWQ_AFF		= 18,

	/* set/get max number of io-wq workers */
	IORING_REGISTER_IOWQ_MAX_WORKERS	= 19,

	/* register/unregister io_uring fd with the ring */
	IORING_REGISTER_RING_FDS		= 20,
	IORING_UNREGISTER_RING_FDS		= 21,

	/* register ring based provide buffer group */
	IORING_REGISTER_PBUF_RING		= 22,
	IORING_UNREGISTER_PBUF_RING		= 23,

	/* sync cancelation API */
	IORING_REGISTER_SYNC_CANCEL		= 24,

	/* register a range of fixed file slots for automatic slot allocation */
	IORING_REGISTER_FILE_ALLOC_RANGE	= 25,

	/* return status information for a buffer group */
	IORING_REGISTER_PBUF_STATUS		= 26,

	/* set/clear busy poll settings */
	IORING_REGISTER_NAPI			= 27,
	IORING_UNREGISTER_NAPI			= 28,

	IORING_REGISTER_CLOCK			= 29,

	/* clone registered buffers from source ring to current ring */
	IORING_REGISTER_CLONE_BUFFERS		= 30,

	/* send MSG_RING without having a ring */
	IORING_REGISTER_SEND_MSG_RING		= 31,

	/* register a netdev hw rx queue for zerocopy */
	IORING_REGISTER_ZCRX_IFQ		= 32,

	/* resize CQ ring */
	IORING_REGISTER_RESIZE_RINGS		= 33,

	IORING_REGISTER_MEM_REGION		= 34,

	/* this goes last */
	IORING_REGISTER_LAST,

	/* flag added to the opcode to use a registered ring fd */
	IORING_REGISTER_USE_REGISTERED_RING	= 1U << 31
};

sqe->flags 支持哪些

/*
 * 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)
/* don't post CQE if request succeeded */
#define IOSQE_CQE_SKIP_SUCCESS	(1U << IOSQE_CQE_SKIP_SUCCESS_BIT)

https://kernel.dk/io_uring.pdf

8.1

When O_DIRECT is used, the kernel must map the application pages into the kernel before it can do IO to them, and subsequently unmap those same pages when IO is done. This can be a costly operation.

这是为什么?

难道不用 O_DIRECT 就不会有问题吗?

难道 aio 不会来 map 这个区域吗? 这个的确很奇怪，因为如果 aio map 了这个区域，那么 aio 就不该有 -EFAULT

write / read 和 aio 也是

迟先生 : io_uring 的接口与实现

更新

https://kernel.dk/io_uring-whatsnew.pdf

io_uring is not an event system

zhihu : io_uring introduction

io_uring 有如此出众的性能，主要来源于以下几个方面：

  1. 用户态和内核态共享提交队列（submission queue）和完成队列（completion queue）
  2. IO 提交和收割可以 offload 给 Kernel，且提交和完成不需要经过系统调用（system call）
  3. 支持 Block 层的 Polling 模式
  4. 通过提前注册用户态内存地址，减少地址映射的开销
  5. 不仅如此，io_uring 还可以完美支持 buffered IO，而 libaio 对于 buffered IO 的支持则一直是被诟病的地方。

Linux I/O 原理和 Zero-copy 技术全面揭秘

各种零拷贝的技术的总结

Efficient IO with io_uring

read the doc : aio’s limitation :

  1. only support async IO for O_DIRECT

    TODO how O_DIRECT works

    1. checked
    2. skip page cache
  2. some constrans for sync submit
  3. api lead noticeable copying

TODO please continue the documentation

read write 会直到事情做完 epoll : 当有的工作做好之后，通知，其实是一个 thread 可以监控一大群的 socket，任何一个 ready 之后就开始工作

https://thenewstack.io/how-io_uring-and-ebpf-will-revolutionize-programming-in-linux/ : 这个也可以参考参考 https://kernel-recipes.org/en/2019/talks/faster-io-through-io_uring/

nice blog provide example code about liburing and directly syscall

[Man page]

io uring and eBPF

细节

IORING_SETUP_NO_SQARRAY

IORING_FEAT_SINGLE_MMAP : 既然 io_uring 和 virtio 为什么都去将 queue 设置为 1 个

到底为什么要 register 来着?

分析下这个报告

AIO not widely used because it only supports one niche use case, even 20 years later.

啊，只有一个 use case ，是什么 use case ?

Requests passing in data structs need to ensure validity only until submit is done, not until completion.

话虽如此，我们知道 request 已经被提交了

Originally io-wq used kernel threads that assumed the identity of the original task when needed. This was risky.

Available in 5.12, io-wq is based on io-threads. These are normal task threads, except they never leave the kernel and they don’t take signals.

什么意思，assumed the identity of original task ?

Native io-threads eliminate security concerns with io-wq offload, for the requests that need that.

It also makes offload a bit more efficient, as no identify switching is needed (files_struct, mm, creds, etc).

It also fixes cases that didn’t previously work, like /proc/self, reading from signalfd, etc.

Enables IORING_SETUP_SQPOLL to work with any file type, or any request in general, and without privilege requirements.

  1. 啊，native worker 可以做到这么多功能!

io-wq used to just block when offloaded.

Normal flow of request is attempt to issue, arm poll if data / space not available.

IORING_RECVSEND_POLL_FIRST Don’t attempt issue first, go straight to poll.

IORING_CQE_F_SOCK_NONEMPTY Previous eg recv() returns if there was more data available.

  1. 什么叫做 arm poll 的问题 ?

TO BE CONTINUE !

即便是这个也是需要读读的

https://man7.org/linux/man-pages/man7/io_uring.7.html

https://lwn.net/Articles/863071/

https://github.com/frevib/io_uring-echo-server

其他

https://news.ycombinator.com/item?id=35547316 : io uring is syscall batch

io uring

Descriptorless files for io_uring

Why you should use io_uring for network I/O

I/O Passthru: Upstreaming a flexible and efficient I/O Path in Linux (fast 2024)

https://www.usenix.org/conference/fast24/presentation/joshi

继续看看这个吧

记住这些东西

https://github.com/axboe/liburing/wiki/io_uring-and-networking-in-2023

有趣的这个东西

https://vmsplice.net/~stefan/stefanha-fosdem-2021.pdf

Redesigned workqueues for io_uring

https://lwn.net/Articles/803070/

本站所有文章转发 CSDN 将按侵权追究法律责任，其它情况随意。