Re: [boost] [serialization] Boost.Serialization 1.33.0BreaksSerializationCode inDLLs

From: martin.ecker_at_[hidden]
Date: 2005-08-16 02:26:53

Hello Robert,

Robert Ramey wrote: 
> I would like to add a couple of new tests to test this kind of thing. 
> You don't have anything suitable for this do you?

Unfortunately, I don't. The only "test case" we have is our entire 
system itself, which is a bit too big I'm afraid.

Basically, it would be beneficial to add a DLL to the test suite
and then write the following test cases:

- Put serialization code in the DLL for a class A. Then serialize an 
instance of class A first from code in the DLL and then from an 
executable that links with the DLL.

- Derive a class B from class A in the DLL. Serialize an instance of 
class B via a base class pointer to class A from code in the DLL and 
from an executable that links with the DLL.

Finally, I'd like to bring your attention to the issue of thread 
safety. Since Boost.Serialization has a number of global objects, 
access to them should be safe-guarded by a synchronization primitive, 
such as a mutex.

This is particularly important for us when dynamically loading DLLs. 
For example, we load and unload DLL plugins during runtime, and we 
have multiple threads that use Boost.Serialization (potentially also 
while one thread is currently loading a DLL). Before I added mutexes 
to all global objects in Boost.Serialization, we had random crashes 
when one thread was dynamically loading a DLL while another thread was 
reading/writing a Boost.Serialization archive. This is because the 
global objects in Boost.Serialization are potentially modified while 
the DLL is being loaded (e.g. if the DLL that is loaded contains 
BOOST_CLASS_EXPORT macros somewhere).

As far as I can tell, the following global objects need to be protected
from simultaneous access from multiple threads:
- tkmap
- ktmap
- void_caster_registry
- iserializer_map
- oserializer_map

Best Regards,
Martin

TAB Austria
Haiderstraße 40
4052 Ansfelden
Austria
Phone: [phone]
Fax: [phone]
E-mail: martin.ecker_at_[hidden]
http://www.tab.at