qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Cliff Jansen" <cliffjan...@gmail.com>
Subject Re: Review Request: Keeps track of Qpid runnable threads and other threads, ensuring that rundown doesn't deadlock.
Date Fri, 01 Jul 2011 19:08:00 GMT


> On 2011-07-01 11:44:51, Steve Huston wrote:
> > My concern with DllMain is that it precludes building qpid as static libraries (and
having this logic still work).

Good point.  I was so focused on the bug as presented I lost complete sight of this valid
use case.

I suppose I could do away with locks and maps altogether with use of TLS storage.  In this
case the only leaked resource would be the TLS slot on multiple uses of LoadLibrary/FreeLibrary,
which is obviously not relevant in the static build case.  So I should be able to use logic
to free the slot in DllMain and comment out that code in a static build.  Does this seem reasonable?


- Cliff


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/987/#review947
-----------------------------------------------------------


On 2011-07-01 03:08:08, Cliff Jansen wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/987/
> -----------------------------------------------------------
> 
> (Updated 2011-07-01 03:08:08)
> 
> 
> Review request for qpid.
> 
> 
> Summary
> -------
> 
> This is the same logic as the preceding version with naming fixes and refinements to
DLL cleanup.
> 
> Cleanup now uses Windows DllMain function to allows cleanup after C++ runtime static
destructors.
> 
> 
> This addresses bug qpid-3256.
>     https://issues.apache.org/jira/browse/qpid-3256
> 
> 
> Diffs
> -----
> 
>   /trunk/qpid/cpp/src/qpid/sys/windows/Thread.cpp 1141687 
> 
> Diff: https://reviews.apache.org/r/987/diff
> 
> 
> Testing
> -------
> 
> Qpid cmake run_tests
> 
> 
> Thanks,
> 
> Cliff
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message