What is the problem with writing singleton in multithreaded enviroment

Thread view:

Ronen Yacov - 14 Jan 2007 08:54 GMT

Hi everybody,

When declaring a singleton, we write:

static CMySingle::instance()
{
static CMySingle instance;
return instance;
}

Why should there be a problem in a multithreaded enviroment?
If one thread creates a static object the other threads will have the
same
instance (becuase its static).

And if there is a problem, how the docuble check locking solves it?

Reply to this Message

EJP - 14 Jan 2007 09:44 GMT

> When declaring a singleton, we write:
>
[quoted text clipped - 3 lines]
> return instance;
> }

Not in Java we don't. Are you in the right newsgroup?

Reply to this Message

Arne Vajhøj - 14 Jan 2007 18:33 GMT

> When declaring a singleton, we write:
>
[quoted text clipped - 10 lines]
>
> And if there is a problem, how the docuble check locking solves it?

That code is C++ not Java.

But anyway. The problem is that the C++ standard mandates that
the constructor can not be called before the first call to
the method. This makes it a non atomic operation.

For some details see
http://blogs.msdn.com/oldnewthing/archive/2004/03/08/85901.aspx !

And double locking does not solve anything. Not in Java, not in C#
and not in C++.

Arne

Reply to this Message

Tom Hawtin - 14 Jan 2007 19:41 GMT

> And double locking does not solve anything. Not in Java, not in C#
> and not in C++.

Double-checked locking? Works fine in Java (1.5+, and in practice 1.4+),
so long as you do it right.

Tom Hawtin

Reply to this Message

Arne Vajhøj - 14 Jan 2007 19:59 GMT

>> And double locking does not solve anything. Not in Java, not in C#
>> and not in C++.
>
> Double-checked locking? Works fine in Java (1.5+, and in practice 1.4+),
> so long as you do it right.

It does not.

It only works in the modified form with the volatile keyword.

Arne

Reply to this Message

Tom Hawtin - 14 Jan 2007 20:10 GMT

>>> And double locking does not solve anything. Not in Java, not in C#
>>> and not in C++.
[quoted text clipped - 3 lines]
>
> It does not.

Does so.

> It only works in the modified form with the volatile keyword.

Yes, correctly implemented it works correctly...

Tom Hawtin

Reply to this Message

Arne Vajhøj - 14 Jan 2007 20:14 GMT

>>>> And double locking does not solve anything. Not in Java, not in C#
>>>> and not in C++.
[quoted text clipped - 9 lines]
>
> Yes, correctly implemented it works correctly...

It is no longer the original double checked locking.

Arne

Reply to this Message

Lew - 14 Jan 2007 20:23 GMT

Arne Vajhøj wrote:

>> And double locking does not solve anything. Not in Java, not in C#
>> and not in C++.

> Double-checked locking? Works fine in Java (1.5+, and in practice
> 1.4+), so long as you do it right.

>> It does not.

> Does so.

>> It only works in the modified form with the volatile keyword

> Yes, correctly implemented it works correctly...

Time out. You guys agree - double-checked locking works, if all restrictions
and limitations are observed. Not if used "by itself", which some might call
"incorrectly", but if used "modified, with volatile", which some might call
"correctly".

- Lew "the Peacemaker"

Reply to this Message

What is the problem with writing singleton in multithreaded enviroment

Free Magazines