Problems with dmalloc5

[Dale Amon]

On Wed, Mar 28, 2012 at 11:24:51AM +1100, Martin Pool wrote:
> Well, I suggest you file a bug against dmalloc on launchpad.net (or
> quote the number if you already have.)  That is generally better than
> just posting them here.
> 
> My next step would be to run this under gdb and find out exactly where
> it is crashing.

Been there, done that. I spent nearly a day on this before
I posted that little program... it started out happening in
a fairly large program :-)
 
> Tangentially, these days I would tend to use Valgrind rather than
> dmalloc.  But, of course, if it's in Ubuntu it ought to work, and I
> suppose you have existing trees that use it.

Yeah, that's the issue. I hate to do changes like that in old
stable code unless I am intending to spend a lot of time 
re-stabilizing and qualifying it afterwards.
 
> I had a brief look,and it turns out you can simplify this program
> quite a lot and it still crashes.  To start with I changed to plain C,
> and then just
> 
> #include        <dmalloc.h>
> 
> int main (int argc, char *argv[]) {
> 	malloc(1);
> 	return 0;
> }

Excellent! I was intending to go back and see if I could 
pull the ObjC and still get the problem.

> I guess dmalloc is out of date with something that changed in libc.

Of course one could note that the copy of dmalloc in Oneiric may
be a bit out of date with the current source of malloc. That might
mean the problem just goes away when precise hits... except I
was not intending on upgrading my laptop to Precise Pangolin
because Ubuntu has that ***** Unity interface which it insists
on clobbering your entire working environment with. Took me
three days to dig out of the hole they put me in when I went
to Oneiric...

So maybe I need to roll my own dmalloc5 from the dmalloc teams
latest sources...