On Tue, May 10, 2022 at 01:13:31PM +0200, Andreas Gruenbacher wrote:
> Kees and Christoph,
>
> thanks for the patch and review.
>
> On Tue, May 10, 2022 at 9:29 AM Christoph Hellwig <h...@infradead.org> wrote:
> > > +/* gfs2_glock_get(), "glock" must be first. */
> > > +struct glock_aspace {
> > > + struct gfs2_glock glock;
> > > + struct address_space mapping;
> > > +};
> >
> > Why does glock need to be first? The whole point of using container_of
> > is that we don't have to make that assumption.
>
> Just needs to be cleaned up in gfs2_glock_get() and
> gfs2_glock_dealloc() as well. I'll do that when applying the patch.
>
> > > static inline struct address_space *gfs2_glock2aspace(struct gfs2_glock
> > > *gl)
> > > {
> > > if (gl->gl_ops->go_flags & GLOF_ASPACE)
> > > - return (struct address_space *)(gl + 1);
> > > + return &(container_of(gl, struct glock_aspace,
> > > glock)->mapping);
> >
> > Do we even need the braces here?
>
> Will use a local variable here, as suggested below.
>
> > > struct inode *inode = mapping->host;
> > > if (mapping->a_ops == &gfs2_meta_aops)
> > > - return (((struct gfs2_glock *)mapping) - 1)->gl_name.ln_sbd;
> > > + return container_of(mapping, struct glock_aspace,
> > > mapping)->glock.gl_name.ln_sbd;
> >
> > A local variable would be really nice for the reader here to decompose
> > this a bit:
> >
> > struct glock_aspace *a =
> > container_of(mapping, struct glock_aspace, mapping);
> >
> > return a->glock.gl_name.ln_sbd;
>
> Yes.
Thanks! So I should leave this with you to arrange, or should I send an
updated patch?
--
Kees Cook
