> [Expanded audience]
>
> This is a generic problem with the present VFS system and also with some
> specific parts of FreeBSD. There is "bleed-over" of all sorts of
> parameters from filesystems and devices that they are mounted on and
> devices that are sourced from them. (e.g. blocksizes etc.)
> luckily at the moment it doesn't usually break things.
>
> I've come across quite a few examples of it though through the
> system at various times.
>
> Has anyone any plans for cleaning up this sort of thing?
> I've been tempted several times.
I remember that Terry had some VFS patches a while back that did
part of this.
And Julian had some SLICE code a while back that could be aboused
to do another part of this.
And I think that if everything was labelled as "vinum" under the
volume management, a lot of this would be abstracted.
> > Bmsafemap structures hang only from buffers associated with
> > filesystem block devices. In looking at the code I can imagine
> > that these could show up if you were running soft updates on
> > the filesystem that contains the block devices (e.g., the root
> > filesystem). On that filesystem, the block device will be
> > encountered during the walk of the files and potentially before
> > the rest of the files associated with its filesystem have been
> > sync'ed to disk. Can you verify that you are running with soft
> > updates on your root filesystem. If so, does turning them off
> > on that filesystem make the panic go away?
> >
> > Kirk McKusick
Alternately, I think Julian had some devfs code that would move
the devices off the root FS, and thus avoid the problem.
Terry Lambert
te...@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.
To Unsubscribe: send mail to majord...@freebsd.org
with "unsubscribe freebsd-current" in the body of the message
