Re: [PATCH] ceph: Annotate struct ceph_monmap with __counted_by

[Xiubo Li]

On 9/18/23 10:04, Gustavo A. R. Silva wrote:
> On 9/17/23 18:25, Xiubo Li wrote:
> > On 9/16/23 04:15, Kees Cook wrote:
> > > Prepare for the coming implementation by GCC and Clang of the 
> > > __counted_by
> > > attribute. Flexible array members annotated with __counted_by can have
> > > their accesses bounds-checked at run-time checking via 
> > > CONFIG_UBSAN_BOUNDS
> > > (for array indexing) and CONFIG_FORTIFY_SOURCE (for 
> > > strcpy/memcpy-family
> > > functions).
> > >
> > > As found with Coccinelle[1], add __counted_by for struct ceph_monmap.
> > > Additionally, since the element count member must be set before 
> > > accessing
> > > the annotated flexible array member, move its initialization earlier.
> > >
> > > [1] 
> > > https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci
> > >
> > > Cc: Ilya Dryomov <idryomov@xxxxxxxxx>
> > > Cc: Xiubo Li <xiubli@xxxxxxxxxx>
> > > Cc: Jeff Layton <jlayton@xxxxxxxxxx>
> > > Cc: "David S. Miller" <davem@xxxxxxxxxxxxx>
> > > Cc: Eric Dumazet <edumazet@xxxxxxxxxx>
> > > Cc: Jakub Kicinski <kuba@xxxxxxxxxx>
> > > Cc: Paolo Abeni <pabeni@xxxxxxxxxx>
> > > Cc: ceph-devel@xxxxxxxxxxxxxxx
> > > Cc: netdev@xxxxxxxxxxxxxxx
> > > Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > > ---
> > >   include/linux/ceph/mon_client.h | 2 +-
> > >   net/ceph/mon_client.c           | 2 +-
> > >   2 files changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/include/linux/ceph/mon_client.h 
> > > b/include/linux/ceph/mon_client.h
> > > index b658961156a0..7a9a40163c0f 100644
> > > --- a/include/linux/ceph/mon_client.h
> > > +++ b/include/linux/ceph/mon_client.h
> > > @@ -19,7 +19,7 @@ struct ceph_monmap {
> > >       struct ceph_fsid fsid;
> > >       u32 epoch;
> > >       u32 num_mon;
> > > -    struct ceph_entity_inst mon_inst[];
> > > +    struct ceph_entity_inst mon_inst[] __counted_by(num_mon);
> > >   };
> > >   struct ceph_mon_client;
> > > diff --git a/net/ceph/mon_client.c b/net/ceph/mon_client.c
> > > index faabad6603db..f263f7e91a21 100644
> > > --- a/net/ceph/mon_client.c
> > > +++ b/net/ceph/mon_client.c
> > > @@ -1136,6 +1136,7 @@ static int build_initial_monmap(struct 
> > > ceph_mon_client *monc)
> > >                      GFP_KERNEL);
> > >       if (!monc->monmap)
> > >           return -ENOMEM;
> > > +    monc->monmap->num_mon = num_mon;
> > >       for (i = 0; i < num_mon; i++) {
> > >           struct ceph_entity_inst *inst = &monc->monmap->mon_inst[i];
> > > @@ -1147,7 +1148,6 @@ static int build_initial_monmap(struct 
> > > ceph_mon_client *monc)
> > >           inst->name.type = CEPH_ENTITY_TYPE_MON;
> > >           inst->name.num = cpu_to_le64(i);
> > >       }
> > > -    monc->monmap->num_mon = num_mon;
> >
> > BTW, is this change related ?
>
> Yes, it is, and it's described in the changelog text.
>
> `num_mon` must be updated before the first access to flex-array 
> `mon_inst`.
> Otherwise the compiler cannot properly instrument the code to catch any
> out-of-bounds access to `mon_inst`.
Okay, got it.

Thanks

- Xiubo


> --
> Gustavo
>
> > >       return 0;
> > >   }
> >
> > Else LGTM.
> >
> > Reviewed-by: Xiubo Li <xiubli@xxxxxxxxxx>
> >
> > Thanks!
> >
> > - Xiubo