Re: [PATCH -next] sched: Dec __cfs_bandwith_used in destroy_cfs_bandwidth()

From: Daniel Jordan
Date: Wed Jul 14 2021 - 09:23:09 EST


On Wed, Jul 14, 2021 at 07:20:39PM +0800, Zhang Qiao wrote:
> And i have another thought is that we can hold the
> hotplug lock before calling cfs_bandwidth_usage_dec().
> This way, fewer modifications are involved.
> What do you think about it?

It is fewer lines, but then hotplug lock is taken pointlessly for
!JUMP_LABEL. Not a huge deal in a slow path like this, just a bit lame.
Adding a new variant seems cleaner if more verbose.