Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news1.oakland.edu!vtc.tacom.army.mil!ulowell.uml.edu!europa.chnt.gtegsc.com!cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!andrew.cmu.edu!postman+ From: Matthew.White@cs.cmu.edu Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: The Future of FreeBSD... Date: Fri, 21 Jul 1995 18:52:10 -0400 Organization: Carnegie Mellon, Pittsburgh, PA Lines: 56 Message-ID: <Ak42yOq00ggLBEvg1B@cs.cmu.edu> References: <3uktse$d9c@hal.nt.tuwien.ac.at> <3ulsro$ssl@agate.berkeley.edu> <Ek3eKzC00ggL1EveAS@cs.cmu.edu> <3uouj0$gk@news.cloud9.net> NNTP-Posting-Host: andrew.cmu.edu In-Reply-To: <3uouj0$gk@news.cloud9.net> Excerpts from netnews.comp.unix.bsd.freebsd.misc: 21-Jul-95 Re: The Future of FreeBSD... Thor Lancelot Simon@clou (1827) > Anyone who believes in this kind of ugly software layering is strongly > encouraged to read Henry Massalin's PhD dissertation, ftp://ftp.cs.columbia.edu/reports/reports-1992/cucs-039-92.ps.Z. While I don't purport to have read the entire paper, I have finished reading through the portion that spoke about kernel structure. I find it largely irrelevant to this discussion. Maybe there is something later in the paper that I have not read yet; I'll read the rest tonight to be sure. From what I have read so far, I have these comments: The flaws in Mach are not necessarily those of the technologies that were developed as part of Mach. The very fact that Mach was a research project here implies the fact that they didn't know what they were doing. They were experimenting, and one of the things they came up with was the idea of a microkernel. The idea of a small piece of hardware dependant code running underneath the rest of the operating system is not isolated in Mach. If Mach's idea of layering didn't work out, then that suggests that something else should be tried, not the whole idea arbitrarily thrown out. Indeed, this is what groups like OSF, IBM and Microsoft have done with their offerings and with much more success (Windows NT will even drive serial ports at 115.2kbps ;-)). The idea of an abstract hardware interface will not go away. This is what originally drove the creation the operating system, later it pushed the development of the idea of the device driver. Each of these inventions was slower than what existed previously, but they added enough value in portability, stability, and ease of use that they were deemed worthwhile. Similarly, the microkernel provides this abstraction for the system hardware. Back to Massalin's paper, I find that it has several interesting ideas that I look forward to reading more about. It is perhaps a too whimsical to be taken seriously by the community at large. Massalin's choice of hardware was also rather unfortunate. By limiting himself to older, CISC, architectures he did not exploit a chance to demonstrate the value of this technology on modern systems. I find the use of assembly language for the implementation of the operating system to be highly questionable. Assembly language is both non-portable and difficult to debug. So while this operating system runs well on the systems for which it has been implemented, it is highly unlikely that it will make its way to other platforms. It is unclear to me at this point whether the assembly implementation is integral to the concepts developed or if perhaps the idea of runtime code generation could be implemented in some other language, such as C or Dylan. BTW, the URL for the Mach project is: http://www.cs.cmu.edu/afs/cs.cmu.edu/project/mach/public/www/mach.html -Matt