Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 08 May 2001 01:29:16 -0700
From:      Jordan Hubbard <jkh@osd.bsdi.com>
To:        juha@saarinen.org
Cc:        dmmiller@cvzoom.net, msmith@FreeBSD.ORG, kris@obsecurity.org, nuno.mailinglists@pt-quorum.com, guilherme@nortenet.pt, freebsd-stable@FreeBSD.ORG
Subject:   RE: CFLAGS Optimization
Message-ID:  <20010508012916Q.jkh@osd.bsdi.com>
In-Reply-To: <KPECIILENDDLPCNIMLOFOENMCDAA.juha@saarinen.org>
References:  <3AF7AA9E.C1DB678A@cvzoom.net> <KPECIILENDDLPCNIMLOFOENMCDAA.juha@saarinen.org>

next in thread | previous in thread | raw e-mail | index | archive | help
From: "Juha Saarinen" <juha@saarinen.org>
Subject: RE: CFLAGS Optimization
Date: Tue, 8 May 2001 20:18:32 +1200

> :: Basically, optimization levels above -O are not supported, so if you
> :: have problems, it's not our fault. 8-)  Basically, stay away from
> :: optimizations with mission-critical stuff such as the kernel and
> :: userland.  For ports, it's OK.
> 
> Even though the kernel and userland might be the bits that would benefit the
> most from optimisations?

You seem to be arguing from an odd perspective.  Yes, gcc has
optimizer bugs.  Yes, you can trigger them easily at higher levels of
optimization.  Yes, it's even conceivable that some bits might benefit
from higher optimization if it were also possible to invoke it more
reliably.  We know all of those things already, so what's the real
question?  Is the question "why aren't they fixed?"  The easy and
obvious answer to that is because nobody has fixed them and you should
certainly feel free to dive on in since they obviously won't fix
themselves and have been around for awhile.  Is the question "should
they be fixed?"  Certainly, but see the previous Q&A.

The next place this subject usually goes is "why don't we use
something other than gcc?"  Any and all actual attempts to do so is
left as an exercise for the reader.  Like riding a unicycle, it's not
as easy as it looks.

- Jordan

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010508012916Q.jkh>