Re: [PATCH 2.6.17-rc6 7/9] Remove some of the kmemleak false positives

["Pekka Enberg" <penberg@cs.helsinki.fi>]

On 13/06/06, Pekka J Enberg <penberg@cs.helsinki.fi> wrote:
> > As far as I understood, Ingo is worried about:
> >
> >         struct s { /* some fields */; char *buf; };
> >
> >         struct s *p = kmalloc(sizeof(struct s) + BUF_SIZE);
> >         p->buf = p + sizeof(struct s);
> >
> > Which could lead to false negative due to p->buf pointing to p.  However,
> > for us to even _find_ p->buf, we would need an incoming pointer _to_ p
> > which makes me think this is not a problem in practice.  Hmm?

On 6/13/06, Catalin Marinas <catalin.marinas@gmail.com> wrote:
> Not exactly. In the above case, Ingo (and me) is worried about having
> a incoming pointer (from other block) equal to p->buf and therefore
> inside the block allocated with kmalloc.

Ah, right, I overlooked that case. But, is it really a leak? That is,
even though we currently don't have a pointer to the beginning fo the
block, we don't know for sure it was a leak. You're now allowed to do:

    p = kmalloc(...);
    p = p + HDR_SIZE;

    /* ... */

    kfree(p - HDR_SIZE);

Do you think we should ban the above?

                                     Pekka