Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!goanna.cs.rmit.edu.au!news.apana.org.au!cantor.edge.net.au!news.teragen.com.au!news.access.net.au!news.mel.connect.com.au!news.mel.aone.net.au!grumpy.fl.net.au!news.webspan.net!newsfeeds.sol.net!news.maxwell.syr.edu!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!europa.clark.net!dciteleport.com!usenet.logical.net!nntp.uac.net!bonkers!web.nmti.com!peter
From: peter@nmti.com (Peter da Silva)
Newsgroups: comp.os.linux.misc,comp.os.linux.networking,comp.os.linux.setup,comp.unix.bsd.bsdi.misc,comp.unix.bsd.misc
Subject: User-space file systems. (Re: Linux vs BSD)
Supersedes: <5fmsul$g2p@web.nmti.com>
Date: 6 Mar 1997 16:56:42 GMT
Organization: Network/development platform support, NMTI
Lines: 62
Message-ID: <5fmt0a$h12@web.nmti.com>
References: <5e6qd5$ivq@cynic.portal.ca> <5fjek4$gtm@cynic.portal.ca> <5fk1t1$3mq@web.nmti.com> <5fkk34$182@cynic.portal.ca>
NNTP-Posting-Host: sonic.nmti.com
Xref: euryale.cc.adfa.oz.au comp.os.linux.misc:163173 comp.os.linux.networking:71108 comp.os.linux.setup:101336 comp.unix.bsd.bsdi.misc:6241 comp.unix.bsd.misc:2739
In article <5fkk34$182@cynic.portal.ca>,
Curt Sampson <cjs@cynic.portal.ca> wrote:
> In article <5fk1t1$3mq@web.nmti.com>, Peter da Silva <peter@nmti.com> wrote:
> >But, Curt, reading a *local* disk block makes two kernel transitions anyway.
> I don't understand this. Reading a local disk block is done in the
> kernel.
I guess i'm jumping too many steps at once, sorry.
I mean that if a program reads a local disk block, that's two kernel
transitions. If it reads a disk block over NFS that's two kernel transitions
plus a minimum of two network packets. If it reads it over NFS to a user-
space NFS server, that's six kernel transitions plus two network packets.
> >Yes, moving NFS into the kernel is a Good Thing, but it's not obvious to
> >me that it's enough of a Good Thing to make it such a big issue.
> It depends on whether your NFS server is a Pentium Pro or a Sun 3. :-)
Perhaps. Of course if your NFS server is a Sun3 you're not running Linux
anyway, and if you're running Linux for an NFS server and you care at all
about performance you've got at least a 486.
> I'm perhaps a little more neutral on this issue that I make myself
> out to be, though. While I certainly wouldn't re-write the BSD NFS
> server to run as a userland daemon,
Neither would I.
> I would seriously consider
> writing one for userland rather than kernel if I were starting from
> scratch.
Which they were (we can go into the background behind that another time...
I think they were silly, but on the other hand it's bought them a lot of
mindshare to have something that wasn't threatened by the USL-CSRG lawsuit
while it was still going on).
> Ah, but now we're not talking about NFS servers, we're talking
> about clients, which are a different kettle of fish.
I understand that, and Linux hasn't got a generic userland file system
interface anyway.
I'm just sorta wary of the whole "Ick, userland NFS" thing, because it
seems to turn people off on userland file systems anyway.
These days the userland transition is not that big a deal. Hell, FreeBSD
uses a userland PPP by default.
--
The Reverend Peter da Silva, ULC, COQO, BOFH.
Har du kramat din varg, idag? `-_-'
--
The Reverend Peter da Silva, ULC, COQO, BOFH.
Har du kramat din varg, idag? `-_-'