From owner-freebsd-stable Fri Mar 24 9:23:16 2000 Delivered-To: freebsd-stable@freebsd.org Received: from screech.weirdnoise.com (209-128-78-198.bayarea.net [209.128.78.198]) by hub.freebsd.org (Postfix) with ESMTP id 70C3437B712; Fri, 24 Mar 2000 09:23:07 -0800 (PST) (envelope-from edhall@screech.weirdnoise.com) Received: from screech.weirdnoise.com (localhost [127.0.0.1]) by screech.weirdnoise.com (8.9.3/8.9.3) with ESMTP id JAA26268; Fri, 24 Mar 2000 09:24:10 -0800 Message-Id: <200003241724.JAA26268@screech.weirdnoise.com> X-Mailer: exmh version 2.0.2 To: Andrew Gallatin Cc: freebsd-alpha@FreeBSD.ORG, freebsd-stable@FreeBSD.ORG Subject: Re: Compiler problems with -O2 (was Re: CVS Trouble, even under 4.0-RELEASE (alpha) HELP!) In-Reply-To: Message from Andrew Gallatin of "Thu, 23 Mar 2000 14:23:12 EST." <14554.28033.439748.801349@grasshopper.cs.duke.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 24 Mar 2000 09:24:10 -0800 From: Ed Hall Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Andrew Gallatin : : You're missing the point almost entirely. FreeBSD's stock gcc -O2 is : demonstrably __broken__ on the alpha. You cannot trust code it : outputs. Interesting. Do we have a test case we can pass along to the gcc folks? Any idea why the Linux folks seem to have better luck with "gcc -O2"? Do they configure gcc differently? (I'm willing to look into it if no one here knows.) I find it interesting that Howard Leadmon's problem with "gcc -O2" showed up as breakage in gcc itself--code that I would assume would be identical to what Linux uses. Specifically, the function check_newline() in /usr/src/contrib/gcc/c-lex.c would seem to be misbehaving. Linux builds gcc with -O2 by default, yet c-lex.c would seem to compile OK there. How are we different? -Ed To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message