Re: Malloc patches, round 3

[Igor Khavkine]

On Wed, Aug 01, 2001 at 01:46:52AM +0300, Kalle Olavi Niemitalo wrote:
> Igor Khavkine <i_khavki@alcor.concordia.ca> writes:
> 
> > --- hurd.orig/ftpfs/node.c  Wed Aug  6 18:08:35 1997
> > +++ hurd/ftpfs/node.c       Tue Jul 31 01:18:31 2001
> > @@ -51,8 +51,10 @@
> >    nn->ncache_next = nn->ncache_prev = 0;
> >  
> >    new = netfs_make_node (nn);
> > -  if (! new)
> > +  if (!new || !nn->rmt_path)
> >      {
> > +      if (nn->rmt_path)
> > +   free (nn->rmt_path);
> >        free (nn);
> >        return ENOMEM;
> >      }
> 
> If nn->rmt_path==NULL and you free(nn), shouldn't you free(new)
> as well?  (netfs_nrele(new) would call netfs_node_norefs and thus
> ihash_locp_remove even though ihash_add hasn't been called.  Is
> that safe?)
> 
> I think it would be clearest not to call netfs_make_node at all
> if the rmt_path allocation fails.
> 
> BTW I would prefer diff --show-c-function.

Very good points. I'm taking all comments under consideration. Also
I will follow Roland's advice and roll a bigger diff, with more
changes of the same kind. But it will take me a bit longer to
come up with it.

Igor