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

[Guillaume Nault <gnault@redhat.com>]

On Thu, Dec 15, 2022 at 02:36:52PM +0000, David Howells wrote:
> 
> Guillaume Nault <gnault@redhat.com> wrote:
> 
> > 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.
> 
> As of this merge window, AF_RXRPC doesn't actually allocate sk_buffs apart
> from when it calls skb_unshare().  It does steal the incoming sk_buffs from
> the UDP socket it uses as a transport, but they're allocated in the IP/IP6
> stack somewhere.
> 
> The UDP transport socket, on the other hand, will allocate sk_buffs for
> transmission, but rxrpc sends an entire UDP packet at a time, each with a
> single sendmsg call.
> 
> Further, this mostly now moved such that the UDP sendmsg calls are performed
> inside an I/O thread.  The application thread does not interact directly with
> the UDP transport socket.
> 
> David

Thanks for the explanations. Looks like we could drop the fs/afs/rxrpc.c
chunk then.