mbox series

[0/4] sched: drop for_each_numa_hop_mask()

Message ID 20230925020528.777578-1-yury.norov@gmail.com (mailing list archive)
Headers show
Series sched: drop for_each_numa_hop_mask() | expand

Message

Yury Norov Sept. 25, 2023, 2:05 a.m. UTC 
Recently added mlx5_cpumask_default_spread() makes for_each_numa_hop_mask()
opencoding cpumask_local_spread().

This series replaces mlx5_cpumask_default_spread() with generic
cpumask_local_spread(). And because mlx5_cpumask_default_spread()
is the only user of for_each_numa_hop_mask() machinery, drops it
entirely.

Regarding for_each_numa_hop_mask(), I've got a small series introducing
a better version of it - for_each_numa_cpu():

https://lore.kernel.org/netdev/20230430171809.124686-1-yury.norov@gmail.com/T/

But with the lack of interest, I believe it's wotrth to drop
for_each_numa_hop_mask() and put for_each_numa_cpu() on hold
until further updates...

Yury Norov (4):
  net: mellanox: drop mlx5_cpumask_default_spread()
  Revert "sched/topology: Introduce for_each_numa_hop_mask()"
  Revert "sched/topology: Introduce sched_numa_hop_mask()"
  lib/cpumask: don't mention for_each_numa_hop_mask in
    cpumask_local_spread()"

 drivers/net/ethernet/mellanox/mlx5/core/eq.c | 28 ++---------------
 include/linux/topology.h                     | 25 ---------------
 kernel/sched/topology.c                      | 33 --------------------
 lib/cpumask.c                                | 21 -------------
 4 files changed, 2 insertions(+), 105 deletions(-)