Commit aebe7ae4 authored by Tejun Heo's avatar Tejun Heo
Browse files

sched_ext: bypass mode shouldn't depend on ops.select_cpu() 



Bypass mode was depending on ops.select_cpu() which can't be trusted as with
the rest of the BPF scheduler. Always enable and use scx_select_cpu_dfl() in
bypass mode.

Signed-off-by: default avatarTejun Heo <tj@kernel.org>
Acked-by: default avatarDavid Vernet <void@manifault.com>
parent cc3e1cac

kernel/sched/ext.c

+15 −13
Original line number Diff line number Diff line
@@ -3126,7 +3126,7 @@ static int select_task_rq_scx(struct task_struct *p, int prev_cpu, int wake_flag 
	if (unlikely(wake_flags & WF_EXEC))

		return prev_cpu;



	if (SCX_HAS_OP(select_cpu)) {

	if (SCX_HAS_OP(select_cpu) && !scx_rq_bypassing(task_rq(p))) {

		s32 cpu;

		struct task_struct **ddsp_taskp;
@@ -3191,7 +3191,7 @@ void __scx_update_idle(struct rq *rq, bool idle) 
{

	int cpu = cpu_of(rq);



	if (SCX_HAS_OP(update_idle)) {

	if (SCX_HAS_OP(update_idle) && !scx_rq_bypassing(rq)) {

		SCX_CALL_OP(SCX_KF_REST, update_idle, cpu_of(rq), idle);

		if (!static_branch_unlikely(&scx_builtin_idle_enabled))

			return;
@@ -4254,21 +4254,23 @@ bool task_should_scx(struct task_struct *p) 
 * the DISABLING state and then cycling the queued tasks through dequeue/enqueue

 * to force global FIFO scheduling.

 *

 * a. ops.enqueue() is ignored and tasks are queued in simple global FIFO order.

 * - ops.select_cpu() is ignored and the default select_cpu() is used.

 *

 * - ops.enqueue() is ignored and tasks are queued in simple global FIFO order.

 *   %SCX_OPS_ENQ_LAST is also ignored.

 *

 * b. ops.dispatch() is ignored.

 * - ops.dispatch() is ignored.

 *

 * c. balance_scx() does not set %SCX_RQ_BAL_KEEP on non-zero slice as slice

 * - balance_scx() does not set %SCX_RQ_BAL_KEEP on non-zero slice as slice

 *   can't be trusted. Whenever a tick triggers, the running task is rotated to

 *   the tail of the queue with core_sched_at touched.

 *

 * d. pick_next_task() suppresses zero slice warning.

 * - pick_next_task() suppresses zero slice warning.

 *

 * e. scx_bpf_kick_cpu() is disabled to avoid irq_work malfunction during PM

 * - scx_bpf_kick_cpu() is disabled to avoid irq_work malfunction during PM

 *   operations.

 *

 * f. scx_prio_less() reverts to the default core_sched_at order.

 * - scx_prio_less() reverts to the default core_sched_at order.

 */

static void scx_ops_bypass(bool bypass)

{
@@ -4338,7 +4340,7 @@ static void scx_ops_bypass(bool bypass) 


		rq_unlock_irqrestore(rq, &rf);



		/* kick to restore ticks */

		/* resched to restore ticks and idle state */

		resched_cpu(cpu);

	}

}