Re: [PATCH v4 04/12] mm/shmem: Support memfile_notifier

[Chao Peng]

On Fri, Feb 11, 2022 at 03:40:09PM -0800, Andy Lutomirski wrote:
> On 1/18/22 05:21, Chao Peng wrote:
> > It maintains a memfile_notifier list in shmem_inode_info structure and
> > implements memfile_pfn_ops callbacks defined by memfile_notifier. It
> > then exposes them to memfile_notifier via
> > shmem_get_memfile_notifier_info.
> > 
> > We use SGP_NOALLOC in shmem_get_lock_pfn since the pages should be
> > allocated by userspace for private memory. If there is no pages
> > allocated at the offset then error should be returned so KVM knows that
> > the memory is not private memory.
> > 
> > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
> 
> >   static int memfile_get_notifier_info(struct inode *inode,
> >                                  struct memfile_notifier_list **list,
> >                                  struct memfile_pfn_ops **ops)
> >   {
> > -   return -EOPNOTSUPP;
> > +   int ret = -EOPNOTSUPP;
> > +#ifdef CONFIG_SHMEM
> > +   ret = shmem_get_memfile_notifier_info(inode, list, ops);
> > +#endif
> > +   return ret;
> >   }
> 
> > +int shmem_get_memfile_notifier_info(struct inode *inode,
> > +                               struct memfile_notifier_list **list,
> > +                               struct memfile_pfn_ops **ops)
> > +{
> > +   struct shmem_inode_info *info;
> > +
> > +   if (!shmem_mapping(inode->i_mapping))
> > +           return -EINVAL;
> > +
> > +   info = SHMEM_I(inode);
> > +   *list = &info->memfile_notifiers;
> > +   if (ops)
> > +           *ops = &shmem_pfn_ops;
> > +
> > +   return 0;
> 
> I can't wrap my head around exactly who is supposed to call these functions
> and when, but there appears to be a missing check that the inode is actually
> a shmem inode.
> 
> What is this code trying to do?  It's very abstract.

This is to be called by memfile_(un)register_notifier in patch-03 to
allow shmem to be connected to memfile_notifer. But as Mike pointed out,
probably introducing a memfile_notifier_register_backing_store() sounds
better so backing store (e.g. shmem) can register itself to
memfile_notifier.

Chao