Date: Tue, 13 May 1997 11:32:06 +0100 (BST) From: Doug Rabson <dfr@nlsystems.com> To: Christoph Kukulies <kuku@gilberto.physik.RWTH-Aachen.DE> Cc: freebsd-current@freefall.FreeBSD.org Subject: Re: weirdo NFS (?) ld problem ? Message-ID: <Pine.BSF.3.95q.970513112927.322K-100000@herring.nlsystems.com> In-Reply-To: <199705131026.MAA03900@gil.physik.rwth-aachen.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 13 May 1997, Christoph Kukulies wrote: > > On Tue, 13 May 1997, Christoph Kukulies wrote: > > > > > > > > I upgraded our NIS/YP NFS cluster of FreeBSD machines to 3.0-current > > > recently (to have real world test conditions :). > > > > > > One machine acts as NIS and NFS server. When one of my 'power' users is > > > compiling and linking a big physics program on the nfs client > > > (The executable is 8.5 MB in size) and he's starting it for the first time > > > he gets a Bus error. Starting it a second time he gets > > > the normal behaviour like printing the startup screen of the program. > > > > > > When he does the same thing on the server the program behaves fine. > > > > > > The client is a PPRO/200. The server is a P5/150. The directories the > > > program sources,objects and binaries reside in are in a mounted > > > /home tree. So I'm suspecting a NFS/VM/cache problem. > > I could reproduce it on a 486 client in the same configuration. > Same picture. > > > > > I am pretty sure there is at least one NFS/VM bug since there are plenty > > of PRs about mmap not working properly in NFS. I have been putting off > > looking for this one since the VM system scares me. I'll have to get to > > it soon though :-(. > > OK. Fine. If you need my environment for reproducing the problem > I could either give you access to my machine or send you the > tree with the program to reproduce the problem. I don't know whether ld > uses mmap. And it seems to occur always the first time after the > binary image has been built (ld). Or where do you think mmap comes > into the game? Having the tree for the program would probably be helpful. I generally debug things here by using gdb across a serial line to my scratch machine, so it would be useful to reproduce it on that machine. I can't promise I will look at it straight away though. I don't think ld uses mmap but it is possible that the mmap problems and the ld problems are caused by the same bug. Its just as likely that there are two bugs though :-(. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 181 951 1891
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95q.970513112927.322K-100000>