From owner-freebsd-hackers Wed Apr 19 11:38:34 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id LAA25790 for hackers-outgoing; Wed, 19 Apr 1995 11:38:34 -0700 Received: from hda.com (hda.com [199.232.40.182]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id LAA25776 for ; Wed, 19 Apr 1995 11:38:28 -0700 Received: (dufault@localhost) by hda.com (8.6.9/8.3) id OAA05851; Wed, 19 Apr 1995 14:37:38 -0400 From: Peter Dufault Message-Id: <199504191837.OAA05851@hda.com> Subject: DEVFS ownership and permissions To: terry@cs.weber.edu (Terry Lambert) Date: Wed, 19 Apr 1995 14:37:38 -0400 (EDT) Cc: hackers@freefall.cdrom.com In-Reply-To: <9504191809.AA19232@cs.weber.edu> from "Terry Lambert" at Apr 19, 95 12:09:27 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1602 Sender: hackers-owner@FreeBSD.org Precedence: bulk Terry Lambert writes: > > > BTW: How the heck do permissions work? > > Ownership and permissions should be specified by the device itself as > part of the registration information. Bill has a raw disk partition he collects his image data on. It is owned by bill and read write by Bill. It is /dev/rsd1e. Where is the persistence of the ownership and the permissions of this raw device which is owned by a regular user? Is it stored on disk someplace and gets established as the system comes up? > Since the device must be there for the device nodes to be valid and > present, and since the devfs is updated by the kernel and can be tied to > kernel events (like load/unload), there is actully no good reason that > this information should be copied to be stored seperately... the only > valid argument is templating of device permissions -- and it is not > very supportable to argue that templating is a better soloution that > simply unloading and reloading the devices (how frequently do you > forsee resetting device permissions?). > > Permissions changed ater that (or ownership changes) can occur in the > device itself. One issue here is in multiple targets for a single > driver; the fact is that for each instance, there must be a name record > generated anyway which will result in a device selector on lookup; the > ownership can be stored as part of that record. I don't understand what you are saying. Peter -- Peter Dufault Real Time Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936 dufault@hda.com Fax: 508 433 5267