Hi.
On 16.11.2020 11:00, Valentin Schneider wrote:
On 15/11/20 22:32, Oleksandr Natalenko wrote:
I'm running v5.10-rc3-rt7 for some time, and I came across this splat
in
dmesg:
```
[118769.951010] ------------[ cut here ]------------
[118769.951013] WARNING: CPU: 19 PID: 146 at kernel/sched/core.c:2013
Err, I didn't pick up on this back then, but isn't that check bogus? If
the
task is enqueued elsewhere, it's valid for it not to be affined
'here'. Also that is_migration_disabled() check within is_cpu_allowed()
makes me think this isn't the best thing to call on a remote task.
---
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 1218f3ce1713..47d5b677585f 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2010,7 +2010,7 @@ static int migration_cpu_stop(void *data)
* valid again. Nothing to do.
*/
if (!pending) {
- WARN_ON_ONCE(!is_cpu_allowed(p, cpu_of(rq)));
+ WARN_ON_ONCE(!cpumask_test_cpu(task_cpu(p), p->cpus_ptr));
goto out;
}
Not sure whether the check is legitimate, but FWIW I've managed to put a
test task [1] (it spawns a lot of threads and applies affinity) into a
permanent unkillable D state here:
```
[<0>] affine_move_task+0x2d3/0x620
[<0>] __set_cpus_allowed_ptr+0x164/0x210
[<0>] sched_setaffinity+0x21a/0x300
[<0>] __x64_sys_sched_setaffinity+0x8c/0xc0
[<0>] do_syscall_64+0x33/0x40
[<0>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
```
I think this corresponds to something around here:
```
$ scripts/faddr2line kernel/sched/core.o affine_move_task+0x2d3
affine_move_task+0x2d3/0x620:
arch_atomic_fetch_sub at
/home/pf/linux-pf-edge/src/linux-5.10/./arch/x86/include/asm/atomic.h:190
(inlined by) atomic_fetch_sub_release at
/home/pf/linux-pf-edge/src/linux-5.10/./include/asm-generic/atomic-instrumented.h:221
(inlined by) __refcount_sub_and_test at
/home/pf/linux-pf-edge/src/linux-5.10/./include/linux/refcount.h:272
(inlined by) __refcount_dec_and_test at
/home/pf/linux-pf-edge/src/linux-5.10/./include/linux/refcount.h:315
(inlined by) refcount_dec_and_test at
/home/pf/linux-pf-edge/src/linux-5.10/./include/linux/refcount.h:333
(inlined by) affine_move_task at
/home/pf/linux-pf-edge/src/linux-5.10/kernel/sched/core.c:2334
```
or:
```
2332 wait_for_completion(&pending->done);
2333
2334 if (refcount_dec_and_test(&pending->refs))
2335 wake_up_var(&pending->refs);
```
I'm not positive about this being directly related to the original
report, but I think it is still worth mentioning.
Thanks.
[1] https://gitlab.com/post-factum/burn_scheduler
--
Oleksandr Natalenko (post-factum)
