On Wed, Aug 03, 2011 at 04:57:42PM +0200, Fabian Groffen wrote:
> On 03-08-2011 16:55:58 +0200, Stefan Manegold wrote:
> > On Wed, Aug 03, 2011 at 04:51:50PM +0200, Fabian Groffen wrote:
> > > On 03-08-2011 16:46:58 +0200, Stefan Manegold wrote:
> > > > Changeset: e680b640d816 for MonetDB
> > > > URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=e680b640d816
> > > > Modified Files:
> > > > monetdb5/mal/mal_stack.c
> > > > Branch: Aug2011
> > > > Log Message:
> > > >
> > > > re-activate "dark code" of function reallocStack() to fix compilation
> > > >
> > > > it appears that mal_box now requires the "dark code" of
> > > > mal_stack's function reallocStack() ...
> > >
> > > Please revert and disable the call to reallocStack. That function is
> > > broken as hell.
> >
> > What is the good replacement for the use of reallocStack() in mal_box, then?
>
> Just don't call the function.
hm, from function depositBox() in monetdb5/mal/mal_box.mx:
[...]
if (i < 0) {
i = newVariable(box->sym, GDKstrdup(name), type);
if (box->val->stksize <= i)
box->val =reallocStack(box->val, STACKINCR);
}
[...]
what happens if the stack is not expanded as required?
> (It returns an alloca'd pointer, or fails with an assertion.)
the current reallocStack() in monetdb5/mal/mal_stack.c uses GDKzalloc(), not
alloca():
MalStkPtr
reallocStack(MalStkPtr s, int cnt)
{
int k;
MalStkPtr old = s;
if (s->stksize > cnt)
return s;
k = ((cnt / STACKINCR) + 1) * STACKINCR;
s = (MalStkPtr) GDKzalloc(stackSize(k));
memcpy(s, old, stackSize(old->stksize));
s->stksize = k;
GDKfree(old);
return s;
}
... does not look too "evil" to me ...
Stefan
> _______________________________________________
> Checkin-list mailing list
> [email protected]
> http://mail.monetdb.org/mailman/listinfo/checkin-list
>
--
| Stefan.Manegold @ CWI.nl | DB Architectures (INS1) |
| http://CWI.nl/~manegold/ | Science Park 123 (L321) |
| Tel.: +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |
_______________________________________________
Checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
