Date: Fri, 05 Dec 2008 11:41:32 +0000 From: Bruce M Simpson <bms@incunabulum.net> To: FreeBSD stable <freebsd-stable@freebsd.org> Cc: Matteo Riondato <matteo@freebsd.org>, Konstantin Belousov <kib@FreeBSD.org> Subject: Issues with mount(8) and FUSE in 7.1-PRERELEASE Message-ID: <4939136C.3000500@incunabulum.net>
next in thread | raw e-mail | index | archive | help
Hi all,
I noticed when porting ext2fuse to run on 7.1-PRERELEASE, that the port
sysutils/fusefs-ntfs needs some additional steps in order to mount
user-space FUSE filesystems on startup. These steps weren't necessary in
6.x.
It seems mount(8) can't deal with new mount binaries unless it has
been explicitly built with support for them; please see
/usr/local/share/doc/ntfs-3g/README.FreeBSD for more info where the
problem is described in more detail.
There is a comment which reads as follows in the mount.c code:
/* XXX: We need to get away from implementing external mount
* programs for every filesystem, and move towards having
* each filesystem properly implement the nmount() system call.
*/
Unfortunately, this assumption seems to be grounded in a very
restricted picture of the filesystem world. FUSE is clearly something
which doesn't fit a limited use case of "let's only mount stuff which
the FreeBSD base system supports", and, to my mind, is applicable to the
vast majority of FreeBSD desktop users.
This seems like an inflexible approach, and I'm curious why this
design choice persists. I understand the technical arguments for
nmount() and I do support them, but not at the expense of functional
utility.
How can we make this work for users out-of-the-box?
thanks,
BMS
P.S. There is a patch by Ale Pulver to change some of this hardcoded
logic, however it doesn't seem complete:
http://people.freebsd.org/~alepulver/current-7.0-mount.diff
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4939136C.3000500>