Re: [PATCH net-next 2/2] KEYS: annotate struct user_key_payload with __counted_by

From: Thorsten Blum

Date: Wed Apr 08 2026 - 08:33:59 EST


On Wed, Apr 08, 2026 at 12:02:25PM +0300, Jarkko Sakkinen wrote:
> On Mon, Apr 06, 2026 at 07:58:10PM +0200, Thorsten Blum wrote:
> > Add the __counted_by() compiler attribute to the flexible array member
> > 'data' to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
> > CONFIG_FORTIFY_SOURCE.
> >
> > Signed-off-by: Thorsten Blum <thorsten.blum@xxxxxxxxx>
> > ---
> > include/keys/user-type.h | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/keys/user-type.h b/include/keys/user-type.h
> > index 386c31432789..2305991f4fcd 100644
> > --- a/include/keys/user-type.h
> > +++ b/include/keys/user-type.h
> > @@ -27,7 +27,8 @@
> > struct user_key_payload {
> > struct rcu_head rcu; /* RCU destructor */
> > unsigned short datalen; /* length of this data */
> > - char data[] __aligned(__alignof__(u64)); /* actual data */
> > + char data[] /* actual data */
> > + __aligned(__alignof__(u64)) __counted_by(datalen);
> > };
> >
> > extern struct key_type key_type_user;
>
> You don't provide any evidence of any improvement.

It's a proactive hardening change to help avoid future mistakes.

The __counted_by() annotation makes the bounds visible to the compiler
and at runtime so that future ->data accesses can be checked against
->datalen.

The current code is correct regarding ->data accesses and doesn't
require any changes.