netdev - Re: [PATCH net-next] udp: introduce and use indirect call wrapper for data ready()

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Mon, 17 Jul 2023 10:47:18 -0400
From: Willem de Bruijn <willemdebruijn.kernel@...il.com>
To: Paolo Abeni <pabeni@...hat.com>, 
 Willem de Bruijn <willemdebruijn.kernel@...il.com>, 
 netdev@...r.kernel.org
Cc: "David S. Miller" <davem@...emloft.net>, 
 Eric Dumazet <edumazet@...gle.com>, 
 Jakub Kicinski <kuba@...nel.org>, 
 David Ahern <dsahern@...nel.org>
Subject: Re: [PATCH net-next] udp: introduce and use indirect call wrapper for
 data ready()

Paolo Abeni wrote:
> On Mon, 2023-07-17 at 09:44 -0400, Willem de Bruijn wrote:
> > Paolo Abeni wrote:
> > > In most cases UDP sockets use the default data ready callback.
> > > This patch Introduces and uses a specific indirect call wrapper for
> > > such callback to avoid an indirect call in fastpath.
> > > 
> > > The above gives small but measurable performance gain under UDP flood.
> > 
> > Interesting. I recently wrote a patch to add indirect call wrappers
> > around getfrag (ip_generic_getfrag), expecting that to improve  UDP
> > senders. Since it's an indirect call on each send call. Not sent,
> > because I did not see measurable gains, at least with a udp_rr bench.
> > 
> > > Signed-off-by: Paolo Abeni <pabeni@...hat.com>
> > > ---
> > > Note that this helper could be used for TCP, too. I did not send such
> > > patch right away because in my tests the perf delta there is below the
> > > noise level even in RR scenarios and the patch would be a little more
> > > invasive - there are more sk_data_ready() invocation places.
> > > ---
> > >  include/net/sock.h | 4 ++++
> > >  net/ipv4/udp.c     | 2 +-
> > >  2 files changed, 5 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/include/net/sock.h b/include/net/sock.h
> > > index 2eb916d1ff64..1b26dbecdcca 100644
> > > --- a/include/net/sock.h
> > > +++ b/include/net/sock.h
> > > @@ -2947,6 +2947,10 @@ static inline bool sk_dev_equal_l3scope(struct sock *sk, int dif)
> > >  }
> > >  
> > >  void sock_def_readable(struct sock *sk);
> > > +static inline void sk_data_ready(struct sock *sk)
> > > +{
> > > +	INDIRECT_CALL_1(sk->sk_data_ready, sock_def_readable, sk);
> > > +}
> > > 
> > 
> > Why introduce a static inline in the header for this?
> > 
> > To reuse it in other protocols later?
> 
> I originally thought about re-using it even for TCP, but showed no gain
> there. I think/hope there could be other users, and I found the code
> nicer this way ;)

Until there are other users I disagree. And maybe even then, as this
is a single line function. It's more readable to see the actual code.

That said, no other concerns from me, if no one else objects.