On Wed 2011.03.16 at 12:30 -0600, Theo de Raadt wrote:
> I don't see any warnings. Please show them.
ksh builds with -Wall by default, so these were emitted:
/home/okan/hack/open/ksh/ksh-current/eval.c: In function 'expand':
/home/okan/hack/open/ksh/ksh-current/eval.c:155: warning: 'x.split' may
be used uninitialized in this function
/home/okan/hack/open/ksh/ksh-current/eval.c:155: warning: 'x.str' may be
used uninitialized in this function
/home/okan/hack/open/ksh/ksh-current/eval.c:268: warning: 'slen' may be
used uninitialized in this function
> If gcc is being fooled by loops, then this is because gcc is stupid.
> it you show the warnings we can judge it.
>
> In particular, the slen one looks totally bogus.
>
> > This diff fixes uninitialized variable warnings emitted by gcc 4.2.1
> > on i386/amd64.
> >
> > Index: eval.c
> > ===================================================================
> > RCS file: /cvs/src/bin/ksh/eval.c,v
> > retrieving revision 1.35
> > diff -u -p -r1.35 eval.c
> > --- eval.c 24 Mar 2010 08:27:26 -0000 1.35
> > +++ eval.c 14 Mar 2011 03:05:10 -0000
> > @@ -159,6 +159,8 @@ expand(char *cp, /* input word */
> > int make_magic;
> > size_t len;
> >
> > + x.split = 0;
> > + x.str = NULL;
> > if (cp == NULL)
> > internal_errorf(1, "expand(NULL)");
> > /* for alias, readonly, set, typeset commands */
> > @@ -267,6 +269,7 @@ expand(char *cp, /* input word */
> > int stype;
> > int slen;
> >
> > + slen = -1;
> > sp = strchr(sp, '\0') + 1; /* skip variable */
> > type = varsub(&x, varname, sp, &stype, &slen);
> > if (type < 0) {
