Use of Linux splice command in plain-text mode for performance increase

[TheAlgorythm]

I've read in a blog post about splice.

The cool thing about this is that all of it happens inside the Linux kernel, which means we won't copy a single byte to userspace (where our program runs). Ideally, splice works by remapping pages and does not actually copy any data, which may improve I/O performance.

The author even provided Rust code which was able to get a throughput of 5.9GiB/s (which is a lot as ordinary cat is able to deliver 1.9GiB/s).

extern crate nix;

use std::env;
use std::fs::File;
use std::io;
use std::os::unix::io::AsRawFd;

use nix::fcntl::{splice, SpliceFFlags};
use nix::unistd::pipe;

const BUF_SIZE: usize = 16384;

fn main() {
    for path in env::args().skip(1) {
        let input = File::open(&path).expect(&format!("fcat: {}: No such file or directory", path));
        let (rd, wr) = pipe().unwrap();
        let stdout = io::stdout();
        let _handle = stdout.lock();

        loop {
            let res = splice(
                input.as_raw_fd(),
                None,
                wr,
                None,
                BUF_SIZE,
                SpliceFFlags::empty(),
            ).unwrap();

            if res == 0 {
                // We read 0 bytes from the input,
                // which means we're done copying.
                break;
            }

            let _res = splice(
                rd,
                None,
                stdout.as_raw_fd(),
                None,
                BUF_SIZE,
                SpliceFFlags::empty(),
            ).unwrap();
        }
    }

[sharkdp]

Thank you for the reference.

I know that article. Maybe we should add a link in this ticket: #304? I don't think we need a new one.