While running a simple DL workload (1 DL (12000, 100000, 100000) task
per CPU) on Arm64 & x86 systems I noticed that some of the
SCHED_WARN_ON() in the rq/running bandwidth (bw) functions trigger.
Patch 1/5 contains a proposal to fix this.
Patch 2-5/5 contain various smaller cleanups I discovered while
debugging the actual issue.
Dietmar Eggemann (5):
sched/deadline: Fix double accounting of rq/running bw in
push_dl_task()
sched/deadline: Remove unused int flags from __dequeue_task_dl()
sched/deadline: Use __sub_running_bw() throughout
dl_change_utilization()
sched/deadline: Cleanup on_dl_rq() handling
sched/deadline: Use return value of SCHED_WARN_ON() in bw accounting
kernel/sched/deadline.c | 28 ++++++++++------------------
1 file changed, 10 insertions(+), 18 deletions(-)
--
2.17.1
