Re: [PATCH 0/5] hybrid union filesystem prototype

[Valerie Aurora]

On Thu, Sep 23, 2010 at 03:18:57PM +0200, Jan Engelhardt wrote:
> 
> On ??, Miklos Szeredi wrote:
> On Thursday 2010-09-02 16:25, Jan Engelhardt wrote:
> >>
> >>There ought to be a reason that other implementations offer
> >>doing multiple branches with a single vfsmount.
> >
> >overdesign
> 
> However, those implementations offer changing the branch configuration
> on the fly, e.g. pulling "out" a branch in the middle. With

Seriously.  What is the use case that justifies the complexity of
implementing this feature (correctly)?  We can all think of
theoretical use cases, but which ones are worth the cost?

> "two-branches only" overlayfs, a multi-branch layout would be
> multiple vfsmounts describable by something like "(((a b) c) d)".
> Say you want to remove branch B. (Supposedly) works today in

There's a difference between "works" and "is correct."  Maybe it is
correct; if so, let's see the code review from VFS experts and some
good stress testing.

-VAL

> aufs/unionfs. How do you pull it off with overlayfs? You can't
> umount the (a b) vfsmount because it is used by c.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/