[PATCH] sched: Do not try to replenish from a non deadline tasks
From: Daniel Wagner
Date: Wed Sep 24 2014 - 09:24:31 EST
When a PI mutex is shared between an deadline task and normal task we
might end up trying to replenish from the normal task. In this case neither
dl_runtime, dl_period or dl_deadline are set. replenish_dl_entity() can't do
anything useful.
The following log is created by
[root@test-kvm ~]# echo 1 > /proc/sys/kernel/ftrace_dump_on_oops
[root@test-kvm ~]# trace-cmd start -e sched -e syscalls:*_futex
I also added an additional trace point to de/enqueue_dl_entity which
prints the PID of dl_se and pi_se.
PID 1534 runs as SCHED_OTHER and PID 1535 runs under SCHED_DEADLINE.
[ 654.000276] pthread_-1535 0.... 4435294us : sys_futex(uaddr: 6010a0, op: 86, val: 1, utime: 0, uaddr2: 0, val3: 5ff)
[ 654.000276] pthread_-1535 0d... 4435295us : sched_pi_setprio: comm=pthread_test pid=1535 oldprio=-1 newprio=-1
[ 654.000276] pthread_-1535 0d... 4435295us : sched_dequeue_dl_entity: comm=pthread_test pid=1535 flags=0
[ 654.000276] pthread_-1535 0d... 4435295us : sched_enqueue_dl_entity: comm=pthread_test pid=1535 pi_comm=pthread_test pi_pid=1535 flags=0
[ 654.000276] pthread_-1535 0dN.. 4435295us : sched_pi_setprio: comm=pthread_test pid=1534 oldprio=120 newprio=-1
[ 654.000276] pthread_-1535 0dN.. 4435296us : sched_dequeue_dl_entity: comm=pthread_test pid=1535 flags=0
[ 654.000276] pthread_-1535 0dN.. 4435296us : sched_stat_wait: comm=ksoftirqd/0 pid=3 delay=5043 [ns]
[ 654.000276] pthread_-1535 0d... 4435296us : sched_switch: prev_comm=pthread_test prev_pid=1535 prev_prio=-1 prev_state=S ==> next_comm=ksoftirqd/0 next_pid=3 next_prio=120
[ 654.000276] ksoftirq-3 0d... 4435297us : sched_stat_runtime: comm=ksoftirqd/0 pid=3 runtime=926 [ns] vruntime=2618803465 [ns]
[ 654.000276] ksoftirq-3 0d... 4435297us : sched_stat_wait: comm=sshd pid=1536 delay=5969 [ns]
[ 654.000276] ksoftirq-3 0d... 4435297us : sched_switch: prev_comm=ksoftirqd/0 prev_pid=3 prev_prio=120 prev_state=S ==> next_comm=sshd next_pid=1536 next_prio=120
[ 654.000276] sshd-1536 0d.h. 4435345us : sched_enqueue_dl_entity: comm=pthread_test pid=1534 pi_comm=pthread_test pi_pid=1535 flags=1
[ 654.000276] sshd-1536 0dNh. 4435346us : sched_wakeup: comm=pthread_test pid=1534 prio=-1 success=1 target_cpu=000
[ 654.000276] sshd-1536 0dN.. 4435501us : sched_stat_sleep: comm=ksoftirqd/0 pid=3 delay=48519 [ns]
[ 654.000276] sshd-1536 0dN.. 4435501us : sched_wakeup: comm=ksoftirqd/0 pid=3 prio=120 success=1 target_cpu=000
[ 654.000276] sshd-1536 0dN.. 4435502us : sched_stat_runtime: comm=sshd pid=1536 runtime=48519 [ns] vruntime=9541100 [ns]
[ 654.000276] sshd-1536 0d... 4435502us : sched_switch: prev_comm=sshd prev_pid=1536 prev_prio=120 prev_state=R ==> next_comm=pthread_test next_pid=1534 next_prio=-1
[ 654.000276] pthread_-1534 0.... 4435503us : sys_futex(uaddr: 6010a0, op: 87, val: 0, utime: 0, uaddr2: 6010a0, val3: 5fe)
[ 654.000276] pthread_-1534 0d... 4435504us : sched_enqueue_dl_entity: comm=pthread_test pid=1535 pi_comm=pthread_test pi_pid=1535 flags=1
[ 654.000276] pthread_-1534 0d... 4435504us : sched_wakeup: comm=pthread_test pid=1535 prio=-1 success=1 target_cpu=000
[ 654.000276] pthread_-1534 0d... 4435504us : sched_pi_setprio: comm=pthread_test pid=1534 oldprio=-1 newprio=120
[ 654.000276] pthread_-1534 0d... 4435504us : sched_dequeue_dl_entity: comm=pthread_test pid=1534 flags=0
[ 654.000276] pthread_-1534 0d... 4435505us : sched_enqueue_dl_entity: comm=pthread_test pid=1534 pi_comm=pthread_test pi_pid=1534 flags=8
[ 654.000276] ---------------------------------
[ 654.000276] Modules linked in:
[ 654.000276] CPU: 0 PID: 1534 Comm: pthread_test Not tainted 3.17.0-rc5+ #49
[ 654.000276] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 654.000276] task: ffff88007a1f7380 ti: ffff88007d2b8000 task.ti: ffff88007d2b8000
[ 654.000276] RIP: 0010:[<ffffffff810653f2>] [<ffffffff810653f2>] enqueue_task_dl+0x2b2/0x330
[ 654.000276] RSP: 0018:ffff88007d2bbd08 EFLAGS: 00010046
[ 654.000276] RAX: 0000000000000000 RBX: ffff88007a1f7380 RCX: 0000000000000000
[ 654.000276] RDX: 0000000000000000 RSI: 0000000000000046 RDI: ffff88007d006c00
[ 654.000276] RBP: ffff88007d2bbd38 R08: 0000000000000000 R09: ffff88007a3d3fa4
[ 654.000276] R10: 0000009837497dad R11: 000000000000000d R12: ffff88007a1f7568
[ 654.000276] R13: 0000000000000008 R14: ffff88007a1f7568 R15: ffff88007a1f7568
[ 654.000276] FS: 00007f48db1bc740(0000) GS:ffffffff81e26000(0000) knlGS:0000000000000000
[ 654.000276] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 654.000276] CR2: 00007f1b9095f330 CR3: 0000000076d0a000 CR4: 00000000000006f0
[ 654.000276] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 654.000276] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 654.000276] Stack:
[ 654.000276] ffff88007a1f7380 ffff88007a1f7380 0000000000026e10 ffffffff81e46520
[ 654.000276] ffff88007c8e5820 ffff88007a1f7568 ffff88007d2bbd70 ffffffff81065578
[ 654.000276] ffffffff81e46520 ffff88007a1f7380 ffff88007a1f7568 ffffffff81a0d800
[ 654.000276] Call Trace:
[ 654.000276] [<ffffffff81065578>] update_curr_dl+0x108/0x230
[ 654.000276] [<ffffffff81065859>] dequeue_task_dl+0x19/0x70
[ 654.000276] [<ffffffff8105c605>] dequeue_task+0x55/0x80
[ 654.000276] [<ffffffff8105f849>] rt_mutex_setprio+0x109/0x2c0
[ 654.000276] [<ffffffff8106b5fb>] __rt_mutex_adjust_prio+0x1b/0x40
[ 654.000276] [<ffffffff8183d058>] rt_mutex_unlock+0x68/0x90
[ 654.000276] [<ffffffff81092aa4>] do_futex+0x594/0x950
[ 654.000276] [<ffffffff81092ecc>] SyS_futex+0x6c/0x150
[ 654.000276] [<ffffffff8100edd1>] ? syscall_trace_enter+0x211/0x220
[ 654.000276] [<ffffffff8183e606>] tracesys+0xdc/0xe1
[ 654.000276] Code: 00 00 00 00 00 00 48 89 83 20 02 00 00 eb 9b be 1a 01 00 00 48 c7 c7 cb 4c bd 81 e8 69 a9 fd ff 48 8b 93 e8 01 00 00 eb bd 0f 0b <0f> 0b 48 c7 c7 78 c4 bc 81 31 c0 c6 05 61 5a eb 00 01 e8 47 ca
[ 654.000276] RIP [<ffffffff810653f2>] enqueue_task_dl+0x2b2/0x330
[ 654.000276] RSP <ffff88007d2bbd08>
[ 654.000276] ---[ end trace aed7e30a2c9541d9 ]---
Signed-off-by: Daniel Wagner <daniel.wagner@xxxxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Juri Lelli <juri.lelli@xxxxxxxxx>
---
kernel/sched/deadline.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 2799441..4b3a80f 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -630,7 +630,7 @@ static void update_curr_dl(struct rq *rq)
__dequeue_task_dl(rq, curr, 0);
if (likely(start_dl_timer(dl_se, curr->dl.dl_boosted)))
dl_se->dl_throttled = 1;
- else
+ else if (curr->dl.dl_runtime > 0)
enqueue_task_dl(rq, curr, ENQUEUE_REPLENISH);
if (!is_leftmost(curr, &rq->dl))
--
1.9.3
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/