[#97652] [Ruby master Feature#16746] Endless method definition — mame@...

Issue #16746 has been reported by mame (Yusuke Endoh).

24 messages 2020/04/01

[#97655] [Ruby master Misc#16747] Repository reorganization request — shyouhei@...

Issue #16747 has been reported by shyouhei (Shyouhei Urabe).

12 messages 2020/04/01

[#97745] [Ruby master Bug#16769] Struct.new(..., immutable: true) — takashikkbn@...

Issue #16769 has been reported by k0kubun (Takashi Kokubun).

10 messages 2020/04/08

[#97803] [Ruby master Misc#16775] DevelopersMeeting20200514Japan — mame@...

Issue #16775 has been reported by mame (Yusuke Endoh).

20 messages 2020/04/10

[#97810] [Ruby master Bug#16776] Regression in coverage library — deivid.rodriguez@...

Issue #16776 has been reported by deivid (David Rodr=EDguez).

11 messages 2020/04/10

[#97828] [Ruby master Misc#16778] Should we stop vendoring default gems code? — deivid.rodriguez@...

Issue #16778 has been reported by deivid (David Rodr=EDguez).

37 messages 2020/04/11

[#97878] [Ruby master Feature#16786] Light-weight scheduler for improved concurrency. — samuel@...

Issue #16786 has been reported by ioquatix (Samuel Williams).

72 messages 2020/04/14

[#97893] [Ruby master Bug#16787] [patch] allow Dir.home to work for non-login procs when $HOME not set — salewski@...

Issue #16787 has been reported by salewski (Alan Salewski).

18 messages 2020/04/15

[#97905] [Ruby master Feature#16791] Shortcuts for attributes of Process::Status — 0xfffffff0@...

Issue #16791 has been reported by 0x81000000 (/ /).

10 messages 2020/04/16

[#97907] [Ruby master Bug#16792] Make Mutex held per Fiber instead of per Thread — eregontp@...

Issue #16792 has been reported by Eregon (Benoit Daloze).

9 messages 2020/04/16

[#97989] [Ruby master Misc#16802] Prefer use of RHS assigment in documentation — samuel@...

Issue #16802 has been reported by ioquatix (Samuel Williams).

10 messages 2020/04/21

[#97992] [Ruby master Misc#16803] Discussion: those internal macros reside in public API headers — shyouhei@...

Issue #16803 has been reported by shyouhei (Shyouhei Urabe).

14 messages 2020/04/21

[#98026] [Ruby master Bug#16809] ruby testsuite fails on s390x alpine (musl) with --with-coroutine=copy — ncopa@...

Issue #16809 has been reported by ncopa (Natanael Copa).

11 messages 2020/04/23

[#98034] [Ruby master Feature#16812] Allow slicing arrays with ArithmeticSequence — zverok.offline@...

Issue #16812 has been reported by zverok (Victor Shepelev).

12 messages 2020/04/23

[#98044] [Ruby master Bug#16814] Segmentation fault in GC while running test/ruby/test_fiber.rb on s390x — Rei.Odaira@...

Issue #16814 has been reported by ReiOdaira (Rei Odaira).

14 messages 2020/04/24

[#98059] [Ruby master Bug#16816] Prematurely terminated Enumerator should stay terminated — headius@...

Issue #16816 has been reported by headius (Charles Nutter).

9 messages 2020/04/24

[#98066] [Ruby master Feature#16818] Rename `Range#%` to `Range#/` — sawadatsuyoshi@...

Issue #16818 has been reported by sawa (Tsuyoshi Sawada).

11 messages 2020/04/26

[ruby-core:97907] [Ruby master Bug#16792] Make Mutex held per Fiber instead of per Thread

From: eregontp@...
Date: 2020-04-16 09:01:59 UTC
List: ruby-core #97907
Issue #16792 has been reported by Eregon (Benoit Daloze).

----------------------------------------
Bug #16792: Make Mutex held per Fiber instead of per Thread
https://bugs.ruby-lang.org/issues/16792

* Author: Eregon (Benoit Daloze)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN
----------------------------------------
Currently, Mutex in CRuby is held per Thread.
In JRuby and TruffleRuby, Mutex is held per Fiber (because it's simply easier implementation-wise).

While a user could theoretically notice the difference, it seems extremely uncommon in practice (probably incorrect synchronization).

The usage pattern for a Mutex is using #synchronize or lock+unlock.
Such a pattern protects/surrounds a region of code using some resource, and such a region of code is always on the same Fiber since it's on a given Ruby "stack".

With #16786 it becomes more relevant to have Mutex held Fiber, otherwise Mutex#lock basically would hurt scalability significantly.
This means, if a Fiber does Mutex#lock and it's already held by another Fiber of the same Thread, and the Thread#scheduler is enabled, instead of just raising an error, or preventing switching entirely until #unlock, we would just go to the scheduler and schedule another Fiber (for instance, the one holding that Mutex, or any other ready to be run Fiber).

This is not a new idea and in fact Crystal already does this with its non-blocking Fibers, which is very similar with #16786:
https://github.com/crystal-lang/crystal/blob/612825a53c831ce7d17368c8211342b199ca02ff/src/mutex.cr#L72

Mutex#lock is just like other blocking operations, so let's make it so building on #16786.
I believe it's the natural and intuitive thing to do for Fiber concurrency with a scheduler.

Queue#pop and SizedQueue#push could be other candidates to handle in a similar way.

Here is an early commit to make Mutex held per Fiber, it's quite trivial as you can see:
https://github.com/ruby/ruby/compare/master...eregon:mutex-per-fiber
It passes test-all and test-spec.



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

Unsubscribe: <mailto:[email protected]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>

In This Thread

Prev Next