On Mon, 2019-06-03 at 10:21 -0400, Steve Dickson wrote:
>
> On 5/31/19 11:52 AM, J. Bruce Fields wrote:
> > On Tue, May 28, 2019 at 04:31:15PM -0400, Trond Myklebust wrote:
> > > +char *
> > > +nfsd_path_strip_root(char *pathname)
> > > +{
> > > + const char *dir = nfsd_path_nfsd_rootdir();
> > > + char *ret;
> > > +
> > > + ret = strstr(pathname, dir);
> > > + if (!ret || ret != pathname)
> > > + return pathname;
> >
> > Shouldn't we return NULL or an error or something here? It seems a
> > little strange not to care if the path began with root or not. I
> > guess
> > I need to look at the caller....
> Well pathname will never be NULL... It is returning what is passed
> in,
> but it might be nice to know about the memory failure.
>
Either way, I figure we also want to canonicalise 'dir' before we apply
it, just in case people have amused themselves by composing rootdir
values of the form '/foo/./bar/./../bar/'.
Another patch forthcoming for this.
--
Trond Myklebust
Linux NFS client maintainer, Hammerspace
trond.myklebust@xxxxxxxxxxxxxxx
