On Sun, Jul 17, 2011 at 09:12:29AM -0400, Ted Unangst wrote:
> On Sun, Jul 17, 2011, Alexander Schrijver wrote:
> >> nope, only ssize_t is guaranteed.
> >
> > This is the only way i could think off.
>
> I don't think you will ever be able to get a string longer than
> SSIZE_MAX into memory, but that looks good.
Agreed,
-Otto
>
> >
> > Index: util.c
> > ===================================================================
> > RCS file: /home/alex/scm/cvsync/src/usr.bin/grep/util.c,v
> > retrieving revision 1.41
> > diff -u -r1.41 util.c
> > --- util.c 11 Jul 2011 20:43:21 -0000 1.41
> > +++ util.c 17 Jul 2011 12:01:56 -0000
> > @@ -169,7 +169,13 @@
> > {
> > regmatch_t pmatch;
> > int c, i, r;
> > - int offset;
> > + regoff_t offset;
> > +
> > + /* size_t will be converted to regoff_t. ssize_t is guaranteed to fit
> > + * into regoff_t */
> > + if (l->len > SSIZE_MAX) {
> > + errx(2, "Line is too big to process");
> > + }
> >
> > c = 0;
> > i = 0;
> > @@ -444,7 +450,7 @@
> > #ifdef SMALL
> > return 0;
> > #else
> > - int j;
> > + regoff_t j;
> > int rtrnVal = REG_NOMATCH;
> >
> > pmatch->rm_so = -1;
