On Tue, Mar 24, 2015 at 10:59:20AM +0100, Tobias Hunger wrote:
> On Tue, Mar 24, 2015 at 5:12 AM, Zbigniew Jędrzejewski-Szmek
> <zbys...@in.waw.pl> wrote:
> > This wouldn't work if fstype was NULL. We also have a list of network 
> > filesystems,
> > whic I assume don't require a device, and we can consult this list instead 
> > of including
> > it here. I pushed the patch with those changes.
> 
> Hmmm, you are right. I was thinking about tmpfs, procfs and such where
> the device can be anything, so I did not feel too comfortable to
> include the network filesystems as those usually need to point to some
> specific server or something. Of course that distinction does not make
> too much sense, but I was so focused on my use case that I missed
> that. Sorry.
> 
> > This list contains a lot of old cruft... OTOH, this doesn't hurt, so maybe
> > it's fine.
> >
> > I wonder if we should support "nodev" fstab option. Then we could remove 
> > everything
> > except overlayfs,ramfs,tmpfs from this list and tell people to use nodev 
> > for other
> > cases.
> 
> The best solution would be to just check /proc/filesystems. That way
> there is no user configuration required and the list keeps up to date
> automagically. I can try my hand at parsing that file if you are ok
> with this.
I'm not sure if parsing it buys us anything. It only contains a list of
filesystems for which modules have been loaded, so it cannot be used
to reliably detect the type of a filesystem we will mount in the future.

Zbyszek
_______________________________________________
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/systemd-devel

Reply via email to