On Thu, 2010-01-14 at 09:41 +0100, Jan Friesse wrote:
> Steve,
> first patch  Committed revision 2643.
> 
> But because of malloc race, corosync can still freeze, so it solves nothing.
> 
> I'm really not sure about leak address space (it doesn't leak me anything).
> 
> There is one solution to join thread. Main problem is with global lock
> deadlock. If you will solve this, you will get another segfault from (to
> me) unknown reason.
> 

If a process calls mmap, it must call munmap when it is done with the
memory.  While it may not appear leaked, it is still mapped into the
process address space and could use all the VM available.

Regards
-steve

> Regards,
>   Honza
> 
> Steven Dake wrote:
> > The first attached patch is good for merge.
> > 
> > The removal of the munamap is not suitable.  It will cause corosync to
> > leak address space.  Need to find a different solution that cleans up
> > properly.
> > 
> > Regards
> > -steve
> > 

_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to