[xml] RE: [xmlsec] xmlSecPtrListEnsureSize failing

[Jesse Pelton <jsp PKC com>]

Your recollection is right, but Daniel checked in a change last April that
makes xmlReallocLoc() return NULL if passed NULL.  See
http://cvs.gnome.org/bonsai/cvsview2.cgi?diff_mode=context&whitespace_mode=s
how&file=xmlmemory.c&branch=&root=/cvs/gnome&subdir=gnome-xml&command=DIFF_F
RAMESET&rev1=1.39&rev2=1.40.  CCing the libxml list so Daniel can comment.

> -----Original Message-----
> From: Aleksey Sanin [mailto:aleksey aleksey com] 
> Sent: Thursday, January 22, 2004 1:29 PM
> To: Rob Richards
> Cc: xmlsec aleksey com
> Subject: Re: [xmlsec] xmlSecPtrListEnsureSize failing
>
>
> xmlRealloc() is realloc() eququivalent. And as far as I can remember,
> ANSI C realloc() MUST accept NULL pointer as first argument.
>
> Aleksey
>
>
> Rob Richards wrote on 1/22/2004, 10:00 AM:
> > > Sorry but I don't see this code. What I have is : 
> > >
> > >     if(newSize < gInitialSize) { 
> > >      newSize = gInitialSize; 
> > >     } 
> > >
> > >     newData = (xmlSecPtr*)xmlRealloc(list->data, 
> sizeof(xmlSecPtr) * 
> > newSize); 
> >
> > That's the origional source. I modified it to the following 
> which fixed the 
> > problem for me: 
> >
> > if(newSize < gInitialSize) { 
> >     newSize = gInitialSize; 
> > } 
> >
> > if (list->data) { 
> >     newData = (xmlSecPtr*)xmlRealloc(list->data, 
> sizeof(xmlSecPtr) * 
> > newSize); 
> > } else { 
> >      newData = (xmlSecPtr*)xmlMalloc(sizeof(xmlSecPtr) * newSize); 
> > } 
> >
> > under the origional source, list->data when entering 
> xmlRealloc was NULL, so 
> > the xmlRealloc call would fail as it explcitly tests for 
> NULL before doing 
> > anything. 
> >
> > > > Am I missing something that wouldnt require the above 
> as I see the call 
> > to 
> > > > xmlFree(list->data) in xmlSecPtrListEmpty, but dont see 
> where list->data 
> > is 
> > > > actually alloc'd. 
> > >
> > > Few lines bellow: 
> > >
> > >     list->data = newData; 
> > >     list->max = newSize; 
> >
> > newData was NULL when returned from xmlRealloc, list->data 
> is just set to 
> > NULL (which is what it currently is anyways). 
> > After this it falls into the xmlSecError handler. 
> list->data doesnt ever 
> > seem to have been initialized anywhere and the origional 
> source is expecting 
> > it to not be pointing at a zero'd out memory block. 
> >
> > Rob 
> _______________________________________________
> xmlsec mailing list
> xmlsec aleksey com
> http://www.aleksey.com/mailman/listinfo/xmlsec