Improve push_rt_task() by propagating the double_lock_balance() usage from find_lock_lowest_rq(), thereby reducing the number of cases where we have to assume rq->lock was dropped. When CONFIG_PREEMPT=n, Unnecessary retries frequently happens in a simple test case: Four different rt priority tasks are restricted to run on two cpus.
Thanks for Steven Rostedt's advice. Signed-off-by: Peng Hao <[email protected]> --- kernel/sched/rt.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index 2e2955a..be0fc43 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -1754,7 +1754,7 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq) !task_on_rq_queued(task))) { double_unlock_balance(rq, lowest_rq); - lowest_rq = NULL; + lowest_rq = RETRY_TASK; break; } } @@ -1830,7 +1830,9 @@ static int push_rt_task(struct rq *rq) /* find_lock_lowest_rq locks the rq if found */ lowest_rq = find_lock_lowest_rq(next_task, rq); - if (!lowest_rq) { + if (!lowest_rq) + goto out; + if (lowest_rq == RETRY_TASK) { struct task_struct *task; /* * find_lock_lowest_rq releases rq->lock -- 1.8.3.1
