Re: [RFC PATCH 1/1] fs,ns: allow copying of shm_mnt mount trees

[Snaipe <me@snai.pe>]

On Fri, Jan 30, 2026 at 9:39 AM Johannes Thumshirn
<Johannes.Thumshirn@wdc.com> wrote:
>
> On 1/29/26 6:39 PM, Snaipe wrote:
> > diff --git a/fs/namespace.c b/fs/namespace.c
> > index d82910f33dc4..f51ad2013662 100644
> > --- a/fs/namespace.c
> > +++ b/fs/namespace.c
> > @@ -38,6 +38,9 @@
> >   #include "pnode.h"
> >   #include "internal.h"
> >
> > +/* For checking memfd bind-mounts via shm_mnt */
> > +#include "../mm/internal.h"
> > +
>
> Hi,
>
> I don't want to comment on the rest of the change as I don't know enough
> of the subject but this screams layering violation to me.

Yes -- which is why I was a bit uncomfortable to do this. I tried a
couple of other approaches before this which did not involve including
mm/internal.h and poking at shm_mnt, but they ended up not working.
The only reason I ended up settling on this is that there is precedent
in including mm/internal.h in other parts of the kernel (fs/exec.c
being one example). This may be a bad justification, but if there's a
better way to do what I'm trying to do here I'm gladly taking it.

-- 
Franklin "Snaipe" Mathieu
🝰 https://snai.pe