From owner-freebsd-alpha Sat Apr 17 17:58:21 1999 Delivered-To: freebsd-alpha@freebsd.org Received: from duke.cs.duke.edu (duke.cs.duke.edu [152.3.140.1]) by hub.freebsd.org (Postfix) with ESMTP id D4F2214E3C for ; Sat, 17 Apr 1999 17:58:01 -0700 (PDT) (envelope-from gallatin@cs.duke.edu) Received: from grasshopper.cs.duke.edu (grasshopper.cs.duke.edu [152.3.145.30]) by duke.cs.duke.edu (8.9.1/8.9.1) with ESMTP id UAA10161; Sat, 17 Apr 1999 20:55:36 -0400 (EDT) Received: (from gallatin@localhost) by grasshopper.cs.duke.edu (8.9.3/8.9.1) id UAA18584; Sat, 17 Apr 1999 20:55:27 -0400 (EDT) (envelope-from gallatin@cs.duke.edu) From: Andrew Gallatin MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Sat, 17 Apr 1999 20:55:27 -0400 (EDT) To: Terry Lambert Cc: alpha@FreeBSD.ORG Subject: Re: egcs ready for alpha? In-Reply-To: <199904172332.QAA23509@usr08.primenet.com> References: <14105.4357.904716.780151@grasshopper.cs.duke.edu> <199904172332.QAA23509@usr08.primenet.com> X-Mailer: VM 6.43 under 20.4 "Emerald" XEmacs Lucid Message-ID: <14105.8549.100121.12762@grasshopper.cs.duke.edu> Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Terry Lambert writes: > > > does not like unaligned data accesses. Unfortuantely, a recent commit > > > put a fault handler in that does fixups on unaligned accesses, and the > > > damn this is defaulted to "on", which does a fat lot of good for getting > > > these problems *resolved* instead of *masked*. > > > > Well, had you paid attention to said commit, you'd realize that the > > default behaviour HAS NOT CHANGED. What this commit has added is the > > ability for the administrator to change this behaviour on a global > > basis via a sysctl. (rather than stabbing at it w/ddb) > > > > Really, the fixup/whine/bus-error choice should be made on a > > per-process basis (as in Digital UNIX's uac command). This allows you > > to turn on silent fixups for broken closed-source programs you cannot > > fix.. > > I think you missed the subsequent discussion about the fact that it > was defaulting to zero, and the original author's intention to change > it to default to 1, since he though that what had been committed would > default to 1 automatically. > > If your response is intended to indicate that this *won't* be defaulted > to 1, then I am very, very happy. > Given that there are multiple sysctl's defining the behaviour, I'm not sure what what you mean by zero and one, but I assume you mean you want it to default to machdep.unaligned_sigbus=1? Well, currently it defaults to: machdep.unaligned_print: 1 machdep.unaligned_fix: 1 machdep.unaligned_sigbus: 0 This amounts to what I call 'whiny fixup', which is the behaviour people have come to expect when running on this platform. At least this is what Digital UNIX, NetBSD and Linux default to. All Hidetoshi Shimokawa's commit did was to give you the ability to easily change this default via a sysctl. ------------------------------------------------------------------------------ Andrew Gallatin, Sr Systems Programmer http://www.cs.duke.edu/~gallatin Duke University Email: gallatin@cs.duke.edu Department of Computer Science Phone: (919) 660-6590 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message