On Mon, Jun 18, 2018 at 10:11:13AM +0200, Christoph Hellwig wrote:
> On Sun, Jun 17, 2018 at 01:10:04PM -0700, Dan Williams wrote:
> > I believe kernel behavior regression is a primary concern as now
> > fallocate() and truncate() can randomly fail where they didn't before.
>
> Fail or block forever due to actions of other unprivileged users.
>
> That is a complete no-go.

They will fail "properly" with some error, which is better than current
situation where those unprivileged users can trigger oops.

Thanks