Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1607851
| From | luca abeni <luca.abeni@santannapisa.it> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth |
| Date | 2017-03-23 21:00 +0100 |
| Message-ID | <toh3P-72e-3@gated-at.bofh.it> (permalink) |
| References | <toh3P-72e-5@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
From: Luca Abeni <luca.abeni@santannapisa.it>
Original GRUB tends to reclaim 100% of the CPU time... And this
allows a CPU hog to starve non-deadline tasks.
To address this issue, allow the scheduler to reclaim only a
specified fraction of CPU time.
Signed-off-by: Luca Abeni <luca.abeni@santannapisa.it>
Tested-by: Daniel Bristot de Oliveira <bristot@redhat.com>
---
kernel/sched/core.c | 6 ++++++
kernel/sched/deadline.c | 7 ++++++-
kernel/sched/sched.h | 6 ++++++
3 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 20c62e7..efa88eb 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -6716,6 +6716,12 @@ static void sched_dl_do_global(void)
raw_spin_unlock_irqrestore(&dl_b->lock, flags);
rcu_read_unlock_sched();
+ if (dl_b->bw == -1)
+ cpu_rq(cpu)->dl.deadline_bw_inv = 1 << 8;
+ else
+ cpu_rq(cpu)->dl.deadline_bw_inv =
+ to_ratio(global_rt_runtime(),
+ global_rt_period()) >> 12;
}
}
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 6035311..e964051 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -212,6 +212,11 @@ void init_dl_rq(struct dl_rq *dl_rq)
#else
init_dl_bw(&dl_rq->dl_bw);
#endif
+ if (global_rt_runtime() == RUNTIME_INF)
+ dl_rq->deadline_bw_inv = 1 << 8;
+ else
+ dl_rq->deadline_bw_inv =
+ to_ratio(global_rt_runtime(), global_rt_period()) >> 12;
}
#ifdef CONFIG_SMP
@@ -871,7 +876,7 @@ extern bool sched_rt_bandwidth_account(struct rt_rq *rt_rq);
*/
u64 grub_reclaim(u64 delta, struct rq *rq)
{
- return (delta * rq->dl.running_bw) >> 20;
+ return (delta * rq->dl.running_bw * rq->dl.deadline_bw_inv) >> 20 >> 8;
}
/*
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 57bb79b..141549b 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -565,6 +565,12 @@ struct dl_rq {
* task blocks
*/
u64 running_bw;
+
+ /*
+ * Inverse of the fraction of CPU utilization that can be reclaimed
+ * by the GRUB algorithm.
+ */
+ u64 deadline_bw_inv;
};
#ifdef CONFIG_SMP
--
2.7.4
Back to linux.kernel | Previous | Next — Next in thread | Find similar | Unroll thread
[RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth luca abeni <luca.abeni@santannapisa.it> - 2017-03-23 21:00 +0100
Re: [RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth Peter Zijlstra <peterz@infradead.org> - 2017-03-24 15:10 +0100
Re: [RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth luca abeni <luca.abeni@santannapisa.it> - 2017-03-24 23:00 +0100
Re: [RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth Steven Rostedt <rostedt@goodmis.org> - 2017-03-25 03:40 +0100
Re: [RFC v5 5/9] sched/deadline: do not reclaim the whole CPU bandwidth Peter Zijlstra <peterz@infradead.org> - 2017-03-27 10:40 +0200
csiph-web