[Openais] Re: bugzilla and bug #37

Steven Dake sdake at mvista.com
Thu Sep 9 16:29:52 PDT 2004


On Thu, 2004-09-09 at 08:57, Mark Haverkamp wrote:
> On Wed, 2004-09-08 at 16:55, Steven Dake wrote:
> > Mark
> 
> [...]
> 
> > Also, there is a patch of bug #37 (memory leaks during configuration
> > changes) available.  Could you try it out?
> > 
> > This should also fix an assert when sending large messages from two
> > nodes at the same time.
> > 
> > Thanks
> > -steve
> > 
> 
> One thing that I noticed, looking at the gmi patch. In the functions
> release_reftwo_char and release_reftwo_iovec, the reftwo pointer is set
> to 0xdeadbeaf.  If you want to set the reftwo pointer being passed in
> shouldn't it be passed in as a reftwo** and then set *reftwo =
> 0xdeadbeaf?  Otherwise, isn't the local copy of the pointer just being
> set?
> 
Thanks for pointing that out.

I changed the code so that if the reftwo is ever used after it has been
freed, the code should assert in most cases (unless the memory is
overwritten).

I'll merge in the patch now.

Thanks
-steve

> Mark.




More information about the Openais mailing list