mbox series

[net-next,v3,0/2] net: dev: BYPASS for lockless qdisc

Message ID cover.1553263445.git.pabeni@redhat.com
Headers show
Series net: dev: BYPASS for lockless qdisc | expand

Message

Paolo Abeni March 22, 2019, 3:01 p.m. UTC
This patch series is aimed at improving xmit performances of lockless qdisc
in the uncontended scenario.

After the lockless refactor pfifo_fast can't leverage the BYPASS optimization.
Due to retpolines the overhead for the avoidables enqueue and dequeue operations
has increased and we see measurable regressions.

The first patch introduces the BYPASS code path for lockless qdisc, and the
second one optimizes such path further. Overall this avoids up to 3 indirect
calls per xmit packet. Detailed performance figures are reported in the 2nd
patch.

 v2 -> v3:
  - qdisc_is_empty() has a const argument (Eric)

 v1 -> v2:
  - use really an 'empty' flag instead of 'not_empty', as
    suggested by Eric

Paolo Abeni (2):
  net: sched: add empty status flag for NOLOCK qdisc
  net: dev: introduce support for sch BYPASS for lockless qdisc

 include/net/sch_generic.h | 11 +++++++++++
 net/core/dev.c            |  9 +++++++++
 net/sched/sch_generic.c   |  3 +++
 3 files changed, 23 insertions(+)

Comments

David Miller March 24, 2019, 1:54 a.m. UTC | #1
From: Paolo Abeni <pabeni@redhat.com>
Date: Fri, 22 Mar 2019 16:01:54 +0100

> This patch series is aimed at improving xmit performances of lockless qdisc
> in the uncontended scenario.
> 
> After the lockless refactor pfifo_fast can't leverage the BYPASS optimization.
> Due to retpolines the overhead for the avoidables enqueue and dequeue operations
> has increased and we see measurable regressions.
> 
> The first patch introduces the BYPASS code path for lockless qdisc, and the
> second one optimizes such path further. Overall this avoids up to 3 indirect
> calls per xmit packet. Detailed performance figures are reported in the 2nd
> patch.
> 
>  v2 -> v3:
>   - qdisc_is_empty() has a const argument (Eric)
> 
>  v1 -> v2:
>   - use really an 'empty' flag instead of 'not_empty', as
>     suggested by Eric

Looks great, series applied, thanks Paolo.