Re: [PATCH] nfsd: better conform to setfacl's method for setting missing ACEs

[Jeff Layton]

On Fri, 2023-07-21 at 18:29 +0200, Andreas Grünbacher wrote:
> Am Fr., 21. Juli 2023 um 16:26 Uhr schrieb Jeff Layton <jlayton@xxxxxxxxxx>:
> > Andreas pointed out that the way we're setting missing ACEs doesn't
> > quite conform to what setfacl does. Change it to better conform to
> > how setfacl does this.
> 
> Thanks, this looks reasonable.
> 
> Andreas
> 
> > Cc: Andreas Grünbacher <andreas.gruenbacher@xxxxxxxxx>
> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > ---
> >  fs/nfsd/nfs4acl.c | 15 ++++++++-------
> >  1 file changed, 8 insertions(+), 7 deletions(-)
> > 
> > Chuck, it might be best to fold this into the original patch, if it
> > looks ok.
> > 
> > diff --git a/fs/nfsd/nfs4acl.c b/fs/nfsd/nfs4acl.c
> > index 64e45551d1b6..9ec61bd0e11b 100644
> > --- a/fs/nfsd/nfs4acl.c
> > +++ b/fs/nfsd/nfs4acl.c
> > @@ -742,14 +742,15 @@ static int nfs4_acl_nfsv4_to_posix(struct nfs4_acl *acl,
> >          *  no owner, owning group, or others entry,  a  copy of  the  ACL
> >          *  owner, owning group, or others entry is added to the Default ACL."
> >          *
> > -        * If none of the requisite ACEs were set, and some explicit user or group
> > -        * ACEs were, copy the requisite entries from the effective set.
> > +        * Copy any missing ACEs from the effective set.
> >          */
> > -       if (!default_acl_state.valid &&
> > -           (default_acl_state.users->n || default_acl_state.groups->n)) {
> > -               default_acl_state.owner = effective_acl_state.owner;
> > -               default_acl_state.group = effective_acl_state.group;
> > -               default_acl_state.other = effective_acl_state.other;
> > +       if (default_acl_state.users->n || default_acl_state.groups->n) {

I think we probably need to also do this if any "valid" bits are set.
IOW, if we explicitly set a default entry only for OWNER@, we also need
ACEs for group and other.

I'll send another revision in a bit. This time, I'll just resend an
updated patch of the original instead of a patch on a patch.

Sorry for the churn!

> > +               if (!(default_acl_state.valid & ACL_USER_OBJ))
> > +                       default_acl_state.owner = effective_acl_state.owner;
> > +               if (!(default_acl_state.valid & ACL_GROUP_OBJ))
> > +                       default_acl_state.group = effective_acl_state.group;
> > +               if (!(default_acl_state.valid & ACL_OTHER))
> > +                       default_acl_state.other = effective_acl_state.other;
> >         }
> > 
> >         *pacl = posix_state_to_acl(&effective_acl_state, flags);
> > --
> > 2.41.0
> > 

-- 
Jeff Layton <jlayton@xxxxxxxxxx>