yup, req was for 4k, which caused an easy segfault. Its wouldnt be that easy if there
was real core within that mem overrun.
/gat
btw dont forget the corresponding bmgr->name - its different, but also buggy
Alexander Mai wrote:
> On Mon, Aug 21, 2000 at 07:36:52AM -0400, Uncle George wrote:
> > > The original" increase_needed = (bmgr->atom_size + size) >
> > > bmgr->atom_avail; " produces a boolean. if the difference is more than
> > > 2000, the rest of the algorithm screws up.maybe u wanna say the
> > > following:
> >
> > {
> > increase_needed = (bmgr->atom_size + size) - bmgr->atom_avail;
> > if (increase_needed > 0 )
> > {
> > increase_needed /= 1000;
> > increase_needed++;
> > increase_needed *= 1000;
>
> Looks like this stuff is still likely to contain some misfeatures (also remembering
> your last report).
> I'll commit your change.
> (did you actually hit the limit? At first glance you always get a minimum
> increase of 1000 bytes!?)
>
> --
> Alexander Mai
> [EMAIL PROTECTED]
