Greg Stein <gstein@lyra.org> writes:
> This one probably doesn't make sense. Right after the alloc, we copy over
> the oldsize data. I'm not sure that we need the unused portion to be zeroed.
> IMO, the add'l space in an svn_string really "shouldn't" need to be zeroed,
> but figured it best to ask for other opinions here.
I think you're right -- this is a case where it would be better to
potentially reveal bugs by not zeroing the memory (not to mention
efficiency concerns).
It's reverted.
-K
> On Tue, Oct 03, 2000 at 09:08:46PM -0000, kfogel@tigris.org wrote:
> > User: kfogel
> > Date: 00/10/03 14:08:46
> >
> > Modified: subversion/libsvn_string svn_string.c
> > Log:
> > Change some calls to apr_palloc() to apr_pcalloc() instead, to get
> > zeroed memory. Did this mainly where a structure is being allocated
> > and it wasn't clear that all the structure's fields were being
> > initialized by the caller, and in a few other places. Obviously, this
> > can't break anything, but please feel free to revert in places where
> > you know you don't need zeroed memory.
> >
> > Did not change the NEW(P,T) and NEWARRAY(P,T,N) macros in fs.h, but
> > wonder if that might be a good idea...?
> >
> > Revision Changes Path
> > 1.48 +1 -1 subversion/subversion/libsvn_string/svn_string.c
> >
> > Index: svn_string.c
> > ===================================================================
> > RCS file: /cvs/subversion/subversion/libsvn_string/svn_string.c,v
> > retrieving revision 1.47
> > retrieving revision 1.48
> > diff -u -r1.47 -r1.48
> > --- svn_string.c 2000/10/02 18:29:04 1.47
> > +++ svn_string.c 2000/10/03 21:08:46 1.48
> > @@ -70,7 +70,7 @@
> > already. */
> >
> > /* malloc new area */
> > - new_area = apr_palloc (pool, request);
> > + new_area = apr_pcalloc (pool, request);
> >
> > /* copy data to new area */
> > memcpy (new_area, data, oldsize);
> >
> >
> >
>
> --
> Greg Stein, http://www.lyra.org/
Received on Sat Oct 21 14:36:10 2006