Johannes Kroll wrote:
> On Mo, 2006-07-10 at 09:34 -0700, rhyde@[EMAIL PROTECTED]
wrote:
> > Okay, I've added #define MALLOC_CHECK 1 to the current build (what
will
> > become v1.86) and I'll see if anything shows up on my machines. If
not,
> > you can try the next version on your's and we'll see what happens.
> > Cheers,
> > Randy Hyde
>
>
> I can recommend valgrind (memcheck) for tracking down such problems. It
> runs your program in a kind of virtual machine, similar to bochs, and
> re****ts any errors it finds, with call chains (also off-by-N bugs, etc).
> ccmalloc is more "lightweight", it's basically a malloc/new/delete
> replacement library which also works very good. Both run on Linux.
>
> - J.
I will look into ccmalloc. I'd really like something I can run under
Windows (I have more test programs I can compile for Windows). Worst
case is that I can put together my own malloc tracking functions.
Fortunately, one of the changes made in HLA v1.82 was a complete
switchover from malloc/free to my own malloc2/free2 routines (which do
a little rudimentary checking already), so extending those is a piece
of cake.
Cheers,
Randy Hyde
|
