mbox series

[net-next,0/5] net_sched: sch_fq: round of optimizations

Message ID 20200503025422.219257-1-edumazet@google.com
Headers show
Series net_sched: sch_fq: round of optimizations | expand

Message

Eric Dumazet May 3, 2020, 2:54 a.m. UTC 
This series is focused on better layout of struct fq_flow to
reduce number of cache line misses in fq_enqueue() and dequeue operations.

Eric Dumazet (5):
  net_sched: sch_fq: avoid touching f->next from fq_gc()
  net_sched: sch_fq: change fq_flow size/layout
  net_sched: sch_fq: use bulk freeing in fq_gc()
  net_sched: sch_fq: do not call fq_peek() twice per packet
  net_sched: sch_fq: perform a prefetch() earlier

 net/sched/sch_fq.c | 86 ++++++++++++++++++++++++++--------------------
 1 file changed, 49 insertions(+), 37 deletions(-)