Re: [Boost-bugs] [Boost C++ Libraries] #8730: Race condition in once_block.hpp

Subject: Re: [Boost-bugs] [Boost C++ Libraries] #8730: Race condition in once_block.hpp
From: Boost C++ Libraries (noreply_at_[hidden])
Date: 2013-06-25 19:03:30

#8730: Race condition in once_block.hpp
  Reporter: apolukhin | Owner: andysem
      Type: Bugs | Status: new
 Milestone: To Be Determined | Component: log
   Version: Boost 1.54.0 | Severity: Problem
Resolution: | Keywords: call_once atomics thread log

Comment (by apolukhin):

 Replying to [comment:1 andysem]:
> Which implementation are you referring to?
> > First of all, once_block_flag is not thread safe.
> What makes you think so? It is POD, and all its modifications in the
 library are made in a thread-safe way (given that you use the multi-
 threaded build of the library).

 POD *does not* guarantee atomicity. Take a look at Boost.Atomic library
 and note that even bools and integers are wrapped around in classes and
 specific processor instructions are used to gurantee atomicity. Those
 instructions are *not* generated by compiler when compiler sees POD type,
 because they work much slower than usual ones.

> > keyword static will do bad things in C++03
> I'm well aware of that, and that's the reason why once_block_flag is

 Even if you rewrite it to Boost.Atomic it is not safe to use it that way.
 Back to the example:
 if (!is_foo_inited) {
     new (place_for_foo) foo_type(some parameters);
     is_foo_inited = true;

 Thread no1 checks for `!is_foo_inited`, enters the `if` block and starts
 to execute the placement `new`. At that time thread no2 checks for
 `!is_foo_inited` and enters the `if` block... That is not good, leads to
 memory leaks and undefined behavior!

Ticket URL: <>
Boost C++ Libraries <>
Boost provides free peer-reviewed portable C++ source libraries.

This archive was generated by hypermail 2.1.7 : 2017-02-16 18:50:13 UTC