Re: [PATCH net v3 2/3] Treewide: Stop corrupting socket's task_frag

[Guillaume Nault <gnault@redhat.com>]

On Thu, Dec 15, 2022 at 12:12:42PM +0000, David Howells wrote:
> 
> Benjamin Coddington <bcodding@redhat.com> wrote:
> 
> > diff --git a/fs/afs/rxrpc.c b/fs/afs/rxrpc.c
> > index eccc3cd0cb70..ac75ad18db83 100644
> > --- a/fs/afs/rxrpc.c
> > +++ b/fs/afs/rxrpc.c
> > @@ -46,6 +46,7 @@ int afs_open_socket(struct afs_net *net)
> >  		goto error_1;
> >  
> >  	socket->sk->sk_allocation = GFP_NOFS;
> > +	socket->sk->sk_use_task_frag = false;
> >  
> >  	/* bind the callback manager's address to make this a server socket */
> >  	memset(&srx, 0, sizeof(srx));
> 
> Possibly this should be done in net/rxrpc/local_object.c too?  Or maybe in
> udp_sock_create() or sock_create_kern()?

UDP tunnels typically don't need to set sk_use_task_frag, as they don't
call sk_page_frag(). One exception would be if they called
ip_append_data() (or ip6_append_data()), but none of them seem to do
that (and I can't see any reason why they would).

And net/rxrpc/local_object.c doesn't seems very different in this regard.

Maybe setting sk_use_task_frag in fs/afs/rxrpc.c was overzealous but
I'm not familiar enough with the AF_RXRPC family to tell. If AF_RXRPC
sockets can't call sk_page_frag() and have no reason to do so in the
future, then it should be safe to drop this chunk.

> David
>