From owner-freebsd-current Wed Sep 20 19:30:53 1995 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id TAA06708 for current-outgoing; Wed, 20 Sep 1995 19:30:53 -0700 Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id TAA06675 for ; Wed, 20 Sep 1995 19:30:44 -0700 Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id TAA07859; Wed, 20 Sep 1995 19:28:07 -0700 From: Terry Lambert Message-Id: <199509210228.TAA07859@phaeton.artisoft.com> Subject: Re: 2.2 looking up To: davidg@root.com Date: Wed, 20 Sep 1995 19:28:07 -0700 (MST) Cc: terry@lambert.org, staff@kyklopen.ping.dk, ugen@latte.worldbank.org, current@FreeBSD.org, bugs@ns1.win.net In-Reply-To: <199509210137.SAA04874@corbin.Root.COM> from "David Greenman" at Sep 20, 95 06:37:04 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1242 Sender: owner-current@FreeBSD.org Precedence: bulk > >> No. It is even worse than before. After a while the entire system > >> begins to core dump. None of the staticly linked files seems to > >> generate sig 11. > > > >Copy on write on mmap()'ed files broken? > > No, it has nothing to do with that. There isn't any difference between > shared (dynamic) programs and non-shared (static) programs in terms of how > they are mapped or how copy-on-write works. crt0.o on my box mmap's the shared libraries into the program address space as the first operation, whereas the unshared crt0.o doesn't have that or the extra code for implementing dlopen(). Definite difference between shared and unshared programs. Perhaps ctr0.o has changed for apps linked shared? Or perhaps it hasn't? > The problem is that something is wrong with paging - wrong pages are ending > up in the VM object. > Please try to be patient; the problem will be fixed shortly. No problem for me, anyway. Just airing a theory. It still sounds like a change in mmap() behaviour. Perhaps auto-placement of the region has failed, or some similar problem is showing up. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.