Re: [PATCH v2 2/2] sctp: delay calls to sk_data_ready() as much as possible

[Marcelo Ricardo Leitner]

On Thu, Apr 07, 2016 at 10:05:32AM +0200, Jakub Sitnicki wrote:
> On Wed, Apr 06, 2016 at 07:53 PM CEST, Marcelo Ricardo Leitner 
>  wrote:
> > Currently, the processing of multiple chunks in a single SCTP packet
> > leads to multiple calls to sk_data_ready, causing multiple wake up
> > signals which are costly and doesn't make it wake up any faster.
> >
> > With this patch it will notice that the wake up is pending and will do it
> > before leaving the state machine interpreter, latest place possible to
> > do it realiably and cleanly.
> >
> > Note that sk_data_ready events are not dependent on asocs, unlike waking
> > up writers.
> >
> > Signed-off-by: Marcelo Ricardo Leitner 
> > ---
> 
> [...]
> 
> > diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
> > index 
> > 7fe56d0acabf66cfd8fe29dfdb45f7620b470ac7..e7042f9ce63b0cfca50cae252f51b60b68cb5731
> >  100644
> > --- a/net/sctp/sm_sideeffect.c
> > +++ b/net/sctp/sm_sideeffect.c
> > @@ -1742,6 +1742,11 @@ out:
> > error = sctp_outq_uncork(>outqueue, gfp);
> > } else if (local_cork)
> > error = sctp_outq_uncork(>outqueue, gfp);
> > +
> > +   if (sctp_sk(ep->base.sk)->pending_data_ready) {
> > +   ep->base.sk->sk_data_ready(ep->base.sk);
> > +   sctp_sk(ep->base.sk)->pending_data_ready = 0;
> > +   }
> > return error;
> >  nomem:
> > error = -ENOMEM;
> 
> Would it make sense to introduce a local variable for ep->base.sk (and
> make this function 535+1 lines long ;-)
> 
>   struct sock *sk = ep->base.sk;
> 
> ... like sctp_ulpq_tail_event() does?

I guess so, yes. Same for sctp_sk() cast then. I´ll post a new version
later, thanks.

  Marcelo

Re: [PATCH v2 2/2] sctp: delay calls to sk_data_ready() as much as possible

[Jakub Sitnicki]

On Wed, Apr 06, 2016 at 07:53 PM CEST, Marcelo Ricardo Leitner 
 wrote:
> Currently, the processing of multiple chunks in a single SCTP packet
> leads to multiple calls to sk_data_ready, causing multiple wake up
> signals which are costly and doesn't make it wake up any faster.
>
> With this patch it will notice that the wake up is pending and will do it
> before leaving the state machine interpreter, latest place possible to
> do it realiably and cleanly.
>
> Note that sk_data_ready events are not dependent on asocs, unlike waking
> up writers.
>
> Signed-off-by: Marcelo Ricardo Leitner 
> ---

[...]

> diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
> index 
> 7fe56d0acabf66cfd8fe29dfdb45f7620b470ac7..e7042f9ce63b0cfca50cae252f51b60b68cb5731
>  100644
> --- a/net/sctp/sm_sideeffect.c
> +++ b/net/sctp/sm_sideeffect.c
> @@ -1742,6 +1742,11 @@ out:
>   error = sctp_outq_uncork(>outqueue, gfp);
>   } else if (local_cork)
>   error = sctp_outq_uncork(>outqueue, gfp);
> +
> + if (sctp_sk(ep->base.sk)->pending_data_ready) {
> + ep->base.sk->sk_data_ready(ep->base.sk);
> + sctp_sk(ep->base.sk)->pending_data_ready = 0;
> + }
>   return error;
>  nomem:
>   error = -ENOMEM;

Would it make sense to introduce a local variable for ep->base.sk (and
make this function 535+1 lines long ;-)

  struct sock *sk = ep->base.sk;

... like sctp_ulpq_tail_event() does?

Thanks,
Jakub