Return to BSD News archive
Xref: sserve comp.unix.wizards:31211 comp.unix.bsd:12760 Newsgroups: comp.unix.wizards,comp.unix.bsd Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!uunet!nwnexus!jhgrud!eskimo!scs From: scs@eskimo.com (Steve Summit) Subject: looking for bsd ffs user-mode implementation quote Message-ID: <CEyuAH.IzC@eskimo.com> Organization: none, at the moment Date: Sat, 16 Oct 1993 01:27:43 GMT Lines: 23 I distinctly remember, back when 4.2bsd came out, reading a nice description of the way the new filesystem had been initially developed in user mode, thus being able to make use of user-mode development tools and debuggers, and not having to rebuild and reboot an entire kernel with each pass through the edit/compile/ debug cycle. (They had used a spare disk drive and a special write-through device driver to enable the user-mode code to talk directly to the disk.) I could have sworn I read this in the "Fast File System for UNIX" paper, but I don't seem to have a copy in my archives of the paper I read then, and the text I remember doesn't appear in any of the copies of the paper I can now find (bsd Unix SMM, ACM TOCS). Does anyone remember the text I'm talking about, and can anyone suggest where it might be? (I'm just trying to cite it as a reference; it was a powerful statement on the virtues of user-mode development of code that classically belongs in, or is on its way into, the kernel.) Steve Summit scs@eskimo.com