Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
sched_ext: Warn on task-based SCX op recursion
The kf_tasks[] design assumes task-based SCX ops don't nest - if they
did, kf_tasks[0] would get clobbered. The old scx_kf_allow() WARN_ONCE
caught invalid nesting via kf_mask, but that machinery is gone now.
Add a WARN_ON_ONCE(current->scx.kf_tasks[0]) at the top of each
SCX_CALL_OP_TASK*() macro. Checking kf_tasks[0] alone is sufficient: all
three variants (SCX_CALL_OP_TASK, SCX_CALL_OP_TASK_RET,
SCX_CALL_OP_2TASKS_RET) write to kf_tasks[0], so a non-NULL value at
entry to any of the three means re-entry from somewhere in the family.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Andrea Righi <arighi@nvidia.com>
+6
-1
+6
-1
kernel/sched/ext.c
+6
-1
kernel/sched/ext.c
···
502
502
* held by try_to_wake_up() with rq tracking via scx_rq.in_select_cpu. So if
503
503
* kf_tasks[] is set, @p's scheduler-protected fields are stable.
504
504
*
505
-
* These macros only work for non-nesting ops since kf_tasks[] is not stacked.
505
+
* kf_tasks[] can not stack, so task-based SCX ops must not nest. The
506
+
* WARN_ON_ONCE() in each macro catches a re-entry of any of the three variants
507
+
* while a previous one is still in progress.
506
508
*/
507
509
#define SCX_CALL_OP_TASK(sch, op, rq, task, args...) \
508
510
do { \
511
+
WARN_ON_ONCE(current->scx.kf_tasks[0]); \
509
512
current->scx.kf_tasks[0] = task; \
510
513
SCX_CALL_OP((sch), op, rq, task, ##args); \
511
514
current->scx.kf_tasks[0] = NULL; \
···
517
514
#define SCX_CALL_OP_TASK_RET(sch, op, rq, task, args...) \
518
515
({ \
519
516
__typeof__((sch)->ops.op(task, ##args)) __ret; \
517
+
WARN_ON_ONCE(current->scx.kf_tasks[0]); \
520
518
current->scx.kf_tasks[0] = task; \
521
519
__ret = SCX_CALL_OP_RET((sch), op, rq, task, ##args); \
522
520
current->scx.kf_tasks[0] = NULL; \
···
527
523
#define SCX_CALL_OP_2TASKS_RET(sch, op, rq, task0, task1, args...) \
528
524
({ \
529
525
__typeof__((sch)->ops.op(task0, task1, ##args)) __ret; \
526
+
WARN_ON_ONCE(current->scx.kf_tasks[0]); \
530
527
current->scx.kf_tasks[0] = task0; \
531
528
current->scx.kf_tasks[1] = task1; \
532
529
__ret = SCX_CALL_OP_RET((sch), op, rq, task0, task1, ##args); \