Date: Sat, 7 Dec 1996 23:52:07 +0100 From: Wolfram Schneider <wosch@cs.tu-berlin.de> To: Poul-Henning Kamp <phk@critter.tfs.com> Cc: Paulo Menezes <paulo@isr.uc.pt>, current@freebsd.org Subject: Re: mount_mfs Message-ID: <199612072252.XAA00532@campa.panke.de> In-Reply-To: <9955.849904535@critter.tfs.com> References: <Pine.SUN.3.95.961206202018.314A-100000@pioneer> <9955.849904535@critter.tfs.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Poul-Henning Kamp writes: >>Ok, that is a solution! But what about mfs? I would not like to create a >>/var /etc /tmp for each of the machines. The real question is can mfs be >>used to create an empty ram-disk rw for /tmp? > >Well, mfs has no real speed advantage over any "real" filesystem since >we got the share VM/buffer code, so I generally avoid it. o no update(4) o no synchronus operation for creating/deleting files on disk >Remember that it isn't really a mfs but more like a >swap-partition-fs. It is a user process with contex switching and twice memory-to-memory copying. I think the word 'swap-partition-fs' is confusing. Mfs is a hack and not fully optimized for speed. See 'The Design and Implementation of the 4.4BSD Operating System', page 302-306. Wolfram
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612072252.XAA00532>