[TUHS] UNIX testing / error injection

[Larry McVoy via TUHS]

On Wed, Apr 15, 2026 at 07:54:08AM -0600, Arnold Robbins via TUHS wrote:
> On Tue, Apr 14, 2026, 10:41???PM Adam Koszek via TUHS <tuhs at tuhs.org> wrote:
> >
> > [ .... ]
> >
> > After UNIX was already running many program, were any of the complex
> > pieces of system code ever developed in the user-space first?
> > Filesystems/schedulers/networking are inherently hard to get right, and the
> > compile -> boot -> see kernel fail wasn???t ever fun cycle for me, and it had
> > to be hell on slow machines..
> 
> If I recall correctly, the initial work on the 4.2BSD fast filesystem
> was done in user space.  I'm not sure where I read that, but it might
> be discussed in the TUHS archives.

When srk prototyped extents in UFS, it was in user space.  He basically 
wacked mkfs to not put a rot delay in there, made a file system, made
some files, and mounted it to prove it worked.  He handed that to me
and told me to make it work in the kernel.

It doubled throughput for sequential I/O but didn't have much of an
effect for normal work loads.  A far greater win is still waiting for
someone to do it, do read ahead on stuff in /usr/include or wherever.
-- 
---
Larry McVoy           Retired to fishing          http://www.mcvoy.com/lm/boat