On Fri, Apr 22, 2016 at 11:17:35AM +0200, Didier Kryn wrote:
> Le 21/04/2016 20:38, KatolaZ a écrit :
> >The simple reason is that the very same gcc has compiled at least 90%
> >of the software you are using in alpha4, and who knows how many other
> >millions of packages. So if there were any such bug in gcc, it would
> >have probably been hit by now, just out of pure probability...
> Jaromil, I agree with one of your other posts in which you say
> that changing optimisation options often reveals a bug in the
> source. But for the argument above, well... I'm not sure 90% of the
> millions of packages have been compiled with -O3.
>
> I remember, sometimes in the 80's, when Microware shipped their
> OS9 with their new ANSI C compiler, I had found a bug in the
> compiler, caused by some level of optimisation - it could be
> demonstrated with an ~5 lines example.
>
It was actually me who wrote that comment :) What I mean is that
*normally* a bug that appears only when code is optimised might signal
a problem in memory management/access. There are some cases in which
-O3 might lead to problems, but they seem to be mostly related to
multithreaded applications and to specific issues with floating-point
optimisations, as also explained here:
http://stackoverflow.com/questions/14850593/when-can-i-confidently-compile-program-with-o3/14853616#14853616
My2Cents
KatolaZ
--
[ Enzo Nicosia aka KatolaZ --- GLUG Catania -- Freaknet Medialab ]
[ me [at] katolaz.homeunix.net --
http://katolaz.homeunix.net -- ]
[ GNU/Linux User:#325780/ICQ UIN: #258332181/GPG key ID 0B5F062F ]
[ Fingerprint: 8E59 D6AA 445E FDB4 A153 3D5A 5F20 B3AE 0B5F 062F ]