[ruby-core:87818] [Ruby trunk Feature#13618] [PATCH] auto fiber schedule for rb_wait_for_single_fd and rb_waitpid

[funny.falcon@gmail.com]

Issue #13618 has been updated by funny_falcon (Yura Sokolov).


> It's also going to be more useful for existing code bases like ActionCable, Puma, Async, etc which use their own IO scheduler.

They have their own IO scheduler because ruby had just native threads, which are bad as IO scheduler.
Ok, I'm not totally right: no scheduler will be good enough for everyone.
But I believe, single green scheduler will be good enough for most of things.

> There is no way you can not be compatible with Thread.scheduler if you use standard Ruby IO. Can you give me an example where this isn't true?

No, looks like I'm not confident to answer :-(
Edit: after writing rest of this, I've recognized, that standard Ruby Mutex and Queue will not be compatible with Thread.scheduler.

> Yes, but you pay all the overhead of "it might be a thread", and gain none of the benefits of green threads.

It has benefits both performance and uniformity benefits.
It will be fast, because scheduler still can switch "threads" sitting on a same native thread as fast as Fibers.

Uniformity, because there will be just single set of tools for synchronization: Mutex, ConditionVariable, Queue.
All these tools are needed regardless of "native" vs "green" scheduler.
If these utils will be universal, they will be easily composed together.
Otherwise mix of "native"/"green" threading will become nightmare.
How they will be composed with `Thread.scheduler`?

> Programmer sanity is much more important to me than performance.

That is my stand point too. I believe, less things programmer need to teach, is better.

So there should be:
- use Threads, Mutex, Queue. If you want performance of eventloop, pass `scheduler` parameter to Thread.create.
That is all.

With `Thread.scheduler=` it becomes:
- you may use Threads, Mutex, Queue.
- but if you want performance of eventloop, you need to choose library, that provides scheduler, Mutex, Queue,
  use that library's primitives thorough your code,
  and never mix core Mutex with that library's Mutex, if you occasionally need to use native threads.

Seriously: Ruby will never be that low level language that will gain serious performance through careful
separation of "green" vs "native" thread concepts.

Look at Go (yeah, i've said that, sorry): it were built to be fast practical language.
It has "green threads". But it has no separation "green vs native".
Single option, that digs into that separation, is "runtime.LockThread()" to give the goroutine separate scheduler on separate native thread.
And Go have no non-blocking io visible to user :-O . It is pretty annoying.

Sure, if one want to gain 99.99% of hardware performance, one will not use Go.
She will use C/C++/Rust, will build their own scheduler and event loop.

But if 95% is just ok, than Go is right tool.

Doubtfully there will be so huge performance difference between
   "explicit Thread.scheduler= + that's scheduler synchronization primitives"
vs "standard hybrid Thread with standard hybrid Mutex/Queue".

Sure, hybrid Threads will be much harder to accomplish.
Sure, I could be mistaken entirely.

----------------------------------------
Feature #13618: [PATCH] auto fiber schedule for rb_wait_for_single_fd and rb_waitpid
https://bugs.ruby-lang.org/issues/13618#change-72836

* Author: normalperson (Eric Wong)
* Status: Assigned
* Priority: Normal
* Assignee: normalperson (Eric Wong)
* Target version: 
----------------------------------------
```
auto fiber schedule for rb_wait_for_single_fd and rb_waitpid

Implement automatic Fiber yield and resume when running
rb_wait_for_single_fd and rb_waitpid.

The Ruby API changes for Fiber are named after existing Thread
methods.

main Ruby API:

    Fiber#start -> enable auto-scheduling and run Fiber until it
		   automatically yields (due to EAGAIN/EWOULDBLOCK)

The following behave like their Thread counterparts:

    Fiber.start - Fiber.new + Fiber#start (prelude.rb)
    Fiber#join - run internal scheduler until Fiber is terminated
    Fiber#value - ditto
    Fiber#run - like Fiber#start (prelude.rb)

Right now, it takes over rb_wait_for_single_fd() and
rb_waitpid() function if the running Fiber is auto-enabled
(cont.c::rb_fiber_auto_sched_p)

Changes to existing functions are minimal.

New files (all new structs and relations should be documented):

    iom.h - internal API for the rest of RubyVM (incomplete?)
    iom_internal.h - internal header for iom_(select|epoll|kqueue).h
    iom_epoll.h - epoll-specific pieces
    iom_kqueue.h - kqueue-specific pieces
    iom_select.h - select-specific pieces
    iom_pingable_common.h - common code for iom_(epoll|kqueue).h
    iom_common.h - common footer for iom_(select|epoll|kqueue).h

Changes to existing data structures:

    rb_thread_t.afrunq   - list of fibers to auto-resume
    rb_vm_t.iom          - Ruby I/O Manager (rb_iom_t) :)

Besides rb_iom_t, all the new structs are stack-only and relies
extensively on ccan/list for branch-less, O(1) insert/delete.

As usual, understanding the data structures first should help
you understand the code.

Right now, I reuse some static functions in thread.c,
so thread.c includes iom_(select|epoll|kqueue).h

TODO:

    Hijack other blocking functions (IO.select, ...)

I am using "double" for timeout since it is more convenient for
arithmetic like parts of thread.c.   Most platforms have good FP,
I think.  Also, all "blocking" functions (rb_iom_wait*) will
have timeout support.

./configure gains a new --with-iom=(select|epoll|kqueue) switch

libkqueue:

  libkqueue support is incomplete; corner cases are not handled well:

    1) multiple fibers waiting on the same FD
    2) waiting for both read and write events on the same FD

  Bugfixes to libkqueue may be necessary to support all corner cases.
  Supporting these corner cases for native kqueue was challenging,
  even.  See comments on iom_kqueue.h and iom_epoll.h for
  nuances.

Limitations

Test script I used to download a file from my server:
----8<---
require 'net/http'
require 'uri'
require 'digest/sha1'
require 'fiber'

url = 'http://80x24.org/git-i-forgot-to-pack/objects/pack/pack-97b25a76c03b489d4cbbd85b12d0e1ad28717e55.idx'

uri = URI(url)
use_ssl = "https" == uri.scheme
fibs = 10.times.map do
  Fiber.start do
    cur = Fiber.current.object_id
    # XXX getaddrinfo() and connect() are blocking
    # XXX resolv/replace + connect_nonblock
    Net::HTTP.start(uri.host, uri.port, use_ssl: use_ssl) do |http|
      req = Net::HTTP::Get.new(uri)
      http.request(req) do |res|
    dig = Digest::SHA1.new
    res.read_body do |buf|
      dig.update(buf)
      #warn "#{cur} #{buf.bytesize}\n"
    end
    warn "#{cur} #{dig.hexdigest}\n"
      end
    end
    warn "done\n"
    :done
  end
end

warn "joining #{Time.now}\n"
fibs[-1].join(4)
warn "joined #{Time.now}\n"
all = fibs.dup

warn "1 joined, wait for the rest\n"
until fibs.empty?
  fibs.each(&:join)
  fibs.keep_if(&:alive?)
  warn fibs.inspect
end

p all.map(&:value)

Fiber.new do
  puts 'HI'
end.run.join
```


---Files--------------------------------
0001-auto-fiber-schedule-for-rb_wait_for_single_fd-and-rb.patch (82.8 KB)


-- 
https://bugs.ruby-lang.org/