Xref: utzoo comp.unix.wizards:6326 comp.arch:3274
Path: utzoo!mnetor!uunet!husc6!cmcl2!brl-adm!umd5!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.unix.wizards,comp.arch
Subject: Re: R/O root
Message-ID: <10320@mimsy.UUCP>
Date: 24 Jan 88 05:09:23 GMT
References: <1438@hplabsz.HPL.HP.COM> <93900011@hcx1>
Followup-To: comp.unix.wizards
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 31

[I have redirected followups to comp.unix.wizards.]

>>If the root filesystem is mounted read-only, will pipes continue to work?

In article <93900011@hcx1> garyb@hcx1.SSD.HARRIS.COM writes:
>[Some Unix variants use] a "pipedev" [to define] the filesystem that
>will be used for pipe space (inodes and such).  The BSD systems I have
>seen [these are 4.2BSD and later --Chris] use the kernel socket routines
>to implement pipes.  I know of no Unix that requires that the root
>file system be used for pipe space ....

While those systems that have a pipedev, which include all 2BSDs,
3BSD, 4.0BSD, 4.1BSD, V6, V7, 32V, System III, and some System Vs
[*], in principle (by the very fact that they provide a pipedev)
allow pipes to be stored on a different file system, I suspect any
such usage was rare.  If you make pipedev != rootdev on such a
system, you will not be able to create pipes until the pipe device
is mounted.  On the other hand, this would reduce the activity on
the root, and hence the chance of something untoward occuring
there.

At any rate, when we ran 4.1BSD, we always had pipedev == rootdev.

-----
[*] System V variants would be much easier to name if we just
peeled off the `System V' part of the label and left the *real*
name showing:  `That Vax runs System 2.2 Unix' is more wieldy
than `That Vax runs System V Release 2 Version 2 Unix'.
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@mimsy.umd.edu	Path:	uunet!mimsy!chris