Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.mel.connect.com.au!news.mira.net.au!vic.news.telstra.net!act.news.telstra.net!psgrain!iafrica.com!uct.ac.za!quagga.ru.ac.za!howland.erols.net!www.nntp.primenet.com!nntp.primenet.com!enews.sgi.com!news.mathworks.com!newsfeed.internetmci.com!in1.uu.net!news.artisoft.com!usenet
From: Terry Lambert <terry@lambert.org>
Newsgroups: comp.os.linux.misc,comp.unix.solaris,comp.unix.bsd.freebsd.misc
Subject: Re: Benchmarking different Unix Operating Systems
Date: Fri, 13 Sep 1996 10:52:15 -0700
Organization: Me
Lines: 43
Distribution: inet
Message-ID: <32399F4F.5130395C@lambert.org>
References: <aak2.842008017@Isis.MsState.Edu> <50p41e$1ie@nnrp1.news.primenet.com>
<517sqa$l78@panix.com> <0mBvnq200YUg0k8Bk0@andrew.cmu.edu>
NNTP-Posting-Host: hecate.artisoft.com
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 2.01 (X11; I; Linux 1.1.76 i486)
Xref: euryale.cc.adfa.oz.au comp.os.linux.misc:129064 comp.unix.solaris:82500 comp.unix.bsd.freebsd.misc:27242
G Sumner Hayes wrote:
] bryan@panix.com (Bryan Althaus) writes:
] > G Sumner Hayes (sumner+@CMU.EDU) wrote:
] > What version of Solaris?
] 2.5.1, I believe. It has AFS support (which may or may not require a
] kernel patch; the powers at CMU take care of that sort of thing) and
] SunOS binary support. Otherwise it should be pretty standard.
[ ... panicy Solaris ... ]
Most likely it is the AFS support. When you write a file system
for the kernel multithreaded, SMP-capable, Solaris, there are a
lot of issues with the entrancy locking model which Sun does not
make clear.
Despite my other posting on the dubious advantages of kernel
multithreading without cooperative user space scheduling support,
the topological equivalents of SMP support and real time that
also come with enabling kernel preemption mean that the kernel
environment should be reasonably expected to have these problems
in nearly any modern kernel architecture. Thus, the problem
itself is something your AFS people should have expected and
dealt with in any case (they will soon need to deal with the
same issue if they expect to run as a FreeBSD FS).
I would not be surprised at all that a third party FS would cause
a kernel panic on occasion -- we were a Sun partner working on a
joint project with Sun, and they simply would not provide enough
information on their locking model to enable us to allow reentrancy
of our FS code at the VFS layer.
Have you tried asking the AFS people to set a global entrancy mutex?
There used to be one available for legacy driver support back in
2.3; I'd be surprised if they ripped it out. This would provide
immediate proof of whether or not the problem is intrinsic to
misuse of the kernel environment by AFS.
Terry Lambert
terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.