Re: [PATCH v2 20/21] fuse: use passthrough getattr in setattr suid/sgid handling

On Sat, May 16, 2026 at 3:20 AM Joanne Koong <[email protected]> wrote:
>
> On Fri, May 15, 2026 at 5:53 PM Joanne Koong <[email protected]> wrote:
> >
> > When refreshing i_mode for suid/sgid kill during setattr, use
> > passthrough getattr if the inode has that enabled.
> >
> > Signed-off-by: Joanne Koong <[email protected]>
> > ---
> >  fs/fuse/dir.c | 6 +++++-
> >  1 file changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
> > index b67b3b334e69..fd1b3fd86968 100644
> > --- a/fs/fuse/dir.c
> > +++ b/fs/fuse/dir.c
> > @@ -2524,7 +2524,11 @@ static int fuse_setattr(struct mnt_idmap *idmap, struct dentry *entry,
> >                          * ia_mode calculation may have used stale i_mode.
> >                          * Refresh and recalculate.
> >                          */
> > -                       ret = fuse_do_getattr(idmap, inode, NULL, file);
> > +                       if (fuse_passthrough_op(inode, FUSE_GETATTR))
> > +                               ret = fuse_passthrough_getattr(inode, NULL,
> > +                                                              STATX_MODE, 0);
> > +                       else
> > +                               ret = fuse_do_getattr(idmap, inode, NULL, file);
>
> I left this part untouched from v1 but there was a discussion [1]
> about whether to enforce that setattr must be passed through if
> getattr is passed through [1]. I think there might be some use cases
> though where having only getattr passed through and not setattr is
> useful (eg server wants fast stat() but needs to intercept attribute
> changes for access control or policy enforcement).

Of course there is. I think I was misunderstood.

What I meant was that *specifically* for the getattr for this purpose
of killpriv, it may make sense to ask the server's mode for
calculating the setattr mode for the server.

But this is just a hunch.
I could be wrong and I have no strong feelings either way.

> I'm not sure if you
> still feel it's better to keep it simple with enforcing that
> getattr+setattr rmust always be passed through together, Amir, but if
> so I'm happy to add that in.

No need.

Thanks,
Amir.