Date:   Mon, 21 Jan 2019 17:53:45 +0100
From:   Björn Töpel <bjorn.topel@...il.com>
To:     Paul Menzel <pmenzel@...gen.mpg.de>
Cc:     Björn Töpel <bjorn.topel@...el.com>,
        intel-wired-lan <intel-wired-lan@...ts.osuosl.org>,
        Netdev <netdev@...r.kernel.org>,
        Magnus Karlsson <magnus.karlsson@...il.com>,
        "Karlsson, Magnus" <magnus.karlsson@...el.com>,
        Jesper Dangaard Brouer <brouer@...hat.com>
Subject: Re: [Intel-wired-lan] [PATCH] i40e: replace switch-statement with if-clause

Den mån 21 jan. 2019 kl 17:42 skrev Paul Menzel <pmenzel@...gen.mpg.de>:
>
> Dear Björn,
>
>
> On 01/21/19 17:33, bjorn.topel@...il.com wrote:
> > From: Björn Töpel <bjorn.topel@...el.com>
> >
> > GCC will generate jump tables for switch-statements with more than 5
> > case statements. An entry into the jump table is an indirect call,
> > which means that for CONFIG_RETPOLINE builds, this is rather
> > expensive.
> >
> > This commit replaces the switch-statement that acts on the XDP program
> > result with an if-clause.
>
> Maybe mention the performance improvement already here. I’d also put it
> into the commit message summary. Something like:
>
> > i40e: Speed up retpoline case by using if-clause
>

Yes, this would, indeed, have been a better summary!

> If that jump tables are a common problem, I wonder, why the compiler
> cannot be adapted to generate better performing code or an option passed
> to the compiler.
>

It might make sense to use -fno-jump-tables or a better value for the
case-values-threshold param for the i40e code. However, doing that
would require a much broader testing, since there are a number of
different places where a switch-statement is used. And depending on
the context, a jump table might still be a better option.


Thank you for the comments!
Björn

> > The if-clause was also refactored into a common function that can be
> > used by AF_XDP zero-copy and non-zero-copy code.
> >
> > Performance prior this patch:
> > $ sudo ./xdp_rxq_info --dev enp134s0f0 --action XDP_DROP
> > Running XDP on dev:enp134s0f0 (ifindex:7) action:XDP_DROP options:no_touch
> > XDP stats       CPU     pps         issue-pps
> > XDP-RX CPU      20      18983018    0
> > XDP-RX CPU      total   18983018
> >
> > RXQ stats       RXQ:CPU pps         issue-pps
> > rx_queue_index   20:20  18983012    0
> > rx_queue_index   20:sum 18983012
> >
> > $ sudo ./xdpsock -i enp134s0f0 -q 20 -n 2 -z -r
> >  sock0@...134s0f0:20 rxdrop
> >                 pps         pkts        2.00
> > rx              14,641,496  144,751,092
> > tx              0           0
> >
> > And after:
> > $ sudo ./xdp_rxq_info --dev enp134s0f0 --action XDP_DROP
> > Running XDP on dev:enp134s0f0 (ifindex:7) action:XDP_DROP options:no_touch
> > XDP stats       CPU     pps         issue-pps
> > XDP-RX CPU      20      24000986    0
> > XDP-RX CPU      total   24000986
> >
> > RXQ stats       RXQ:CPU pps         issue-pps
> > rx_queue_index   20:20  24000985    0
> > rx_queue_index   20:sum 24000985
> >
> >   +26%
> >
> > $ sudo ./xdpsock -i enp134s0f0 -q 20 -n 2 -z -r
> >  sock0@...134s0f0:20 rxdrop
> >                 pps         pkts        2.00
> > rx              17,623,578  163,503,263
> > tx              0           0
> >
> >   +20%
> >
> > Signed-off-by: Björn Töpel <bjorn.topel@...el.com>
> > ---
> >  drivers/net/ethernet/intel/i40e/i40e_txrx.c   | 31 ++++---------------
> >  .../ethernet/intel/i40e/i40e_txrx_common.h    | 27 ++++++++++++++++
> >  drivers/net/ethernet/intel/i40e/i40e_xsk.c    | 24 ++------------
> >  3 files changed, 35 insertions(+), 47 deletions(-)
>
> […]
>
>
> Kind regards,
>
> Paul
>
> _______________________________________________
> Intel-wired-lan mailing list
> Intel-wired-lan@...osl.org
> https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

Powered by blists - more mailing lists