On Sat, Apr 28, 2001 at 09:38:02AM +0200, Steve O'Hara-Smith wrote:
> On Sat, 28 Apr 2001 01:27:33 +0200
> "Dimitry Andric" <[EMAIL PROTECTED]> wrote:
>
> DA> Squid with gcc 2.95.2 and optimization (both -O, -O2 and -O666), and
> DA> I can assure you it bombed out with inexplicable null pointer
> DA> accesses. Yet when you compile with -O0, no such thing happens...
>
> I have been working getting swish++ set up as a port (4.3-STABLE) and
> I've found that the search program only works if compiled with no -O setting,
> -O3 (the original) and -O cause segmentation violation, while -O2 gave an
> illegal instruction trap. The index program OTOH appears to work with all
> optimisations settings.
>
> It makes me wonder just how safe -O is :(
>
Although it is quite possible that gcc generates incorrect code in some
cases when invoked with -O it is not very likely.
I would say that it is much more likely that the code which is being
compiled contains a bug that is exposed by the optimization and that the
code just happens to work when comiled with -O0.
Generally I would say that -O is *safer* than -O0 for the simple reason
that it is used more and therefore gets more testing.
--
<Insert your favourite quote here.>
Erik Trulsson
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-stable" in the body of the message
