Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 11 Sep 1995 10:35:20 -0700 (MST)
From:      Terry Lambert <terry@lambert.org>
To:        jkh@time.cdrom.com (Jordan K. Hubbard)
Cc:        current@freefall.freebsd.org
Subject:   Re: Is nullfs broken in -current?
Message-ID:  <199509111735.KAA18641@phaeton.artisoft.com>
In-Reply-To: <199509110722.AAA01160@time.cdrom.com> from "Jordan K. Hubbard" at Sep 11, 95 00:22:14 am

next in thread | previous in thread | raw e-mail | index | archive | help
> I just attempted to use it to loop "/usr/src" and "/usr/obj" mounts on
> my system (^&*$@%!! broken make macros!) and every time I compile
> anything substantive from /usr/src I now get a divide by zero and
> panic.  I could go into more detail here and will if this isn't
> immediately reproducible by those interested in nullfs.  Last person
> I saw in that code was...  David?

The NULL and UNION FS's are problematic.  So is the PORTAL.  The page
management changes will affect both of them in terms of page propagation,
but should not in fact affect other areas significantly.  That is, the
page ops need to be written to pass the pages through to the underlying
layer (layerS, in the union case).

It's getting on time to consider a file system test bed.

The page operations are clearly the cause of the recent instabilities;
they are just as clearly a necessary step forward in terms of performance
and the ability to go multithreaded in support of SMP and multithreading
(this last a significant performance win in even a UP kernel -- to the
tune of a 70% increase in overall file system throuput in the USL UFS
implementation on a UP box, actually).


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199509111735.KAA18641>