Messages in this thread Patch in this message | | | Date | Sun, 17 Nov 2019 11:41:12 +0100 | From | Ingo Molnar <> | Subject | [GIT PULL v2] scheduler fixes |
| |
* Linus Torvalds <torvalds@linux-foundation.org> wrote:
> On Sat, Nov 16, 2019 at 2:44 PM Valentin Schneider > <valentin.schneider@arm.com> wrote: > > > > > Valentin Schneider (2): > > > sched/uclamp: Fix overzealous type replacement > > > > This one got a v2 (was missing one location), acked by Vincent: > > > > 20191115103908.27610-1-valentin.schneider@arm.com > > > > > sched/topology, cpuset: Account for housekeeping CPUs to avoid empty cpumasks > > > > And this one is no longer needed, as Michal & I understood (IOW the fix in > > rc6 is sufficient), see: > > > > c425c5cb-ba8a-e5f6-d91c-5479779cfb7a@arm.com > > Ingo, what do you want me to do? Pull it anyway and send updates > later? Or skip this pull request? > > I'll leave it pending for now,
We ended up zapping the final two commits from sched/urgent.
Please pull the latest sched-urgent-for-linus git tree from:
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git sched-urgent-for-linus
# HEAD: 6e1ff0773f49c7d38e8b4a9df598def6afb9f415 sched/uclamp: Fix incorrect condition
Misc fixes:
- Fix potential deadlock under CONFIG_DEBUG_OBJECTS=y - PELT metrics update ordering fix - uclamp logic fix
Thanks,
Ingo
------------------> Peter Zijlstra (1): sched/core: Avoid spurious lock dependencies
Qais Yousef (1): sched/uclamp: Fix incorrect condition
Vincent Guittot (1): sched/pelt: Fix update of blocked PELT ordering
kernel/sched/core.c | 5 +++-- kernel/sched/fair.c | 29 ++++++++++++++++++++--------- 2 files changed, 23 insertions(+), 11 deletions(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 0f2eb3629070..44123b4d14e8 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -1065,7 +1065,7 @@ uclamp_update_active(struct task_struct *p, enum uclamp_id clamp_id) * affecting a valid clamp bucket, the next time it's enqueued, * it will already see the updated clamp bucket value. */ - if (!p->uclamp[clamp_id].active) { + if (p->uclamp[clamp_id].active) { uclamp_rq_dec_id(rq, p, clamp_id); uclamp_rq_inc_id(rq, p, clamp_id); } @@ -6019,10 +6019,11 @@ void init_idle(struct task_struct *idle, int cpu) struct rq *rq = cpu_rq(cpu); unsigned long flags; + __sched_fork(0, idle); + raw_spin_lock_irqsave(&idle->pi_lock, flags); raw_spin_lock(&rq->lock); - __sched_fork(0, idle); idle->state = TASK_RUNNING; idle->se.exec_start = sched_clock(); idle->flags |= PF_IDLE; diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 22a2fed29054..69a81a5709ff 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7547,6 +7547,19 @@ static void update_blocked_averages(int cpu) rq_lock_irqsave(rq, &rf); update_rq_clock(rq); + /* + * update_cfs_rq_load_avg() can call cpufreq_update_util(). Make sure + * that RT, DL and IRQ signals have been updated before updating CFS. + */ + curr_class = rq->curr->sched_class; + update_rt_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &rt_sched_class); + update_dl_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &dl_sched_class); + update_irq_load_avg(rq, 0); + + /* Don't need periodic decay once load/util_avg are null */ + if (others_have_blocked(rq)) + done = false; + /* * Iterates the task_group tree in a bottom up fashion, see * list_add_leaf_cfs_rq() for details. @@ -7574,14 +7587,6 @@ static void update_blocked_averages(int cpu) done = false; } - curr_class = rq->curr->sched_class; - update_rt_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &rt_sched_class); - update_dl_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &dl_sched_class); - update_irq_load_avg(rq, 0); - /* Don't need periodic decay once load/util_avg are null */ - if (others_have_blocked(rq)) - done = false; - update_blocked_load_status(rq, !done); rq_unlock_irqrestore(rq, &rf); } @@ -7642,12 +7647,18 @@ static inline void update_blocked_averages(int cpu) rq_lock_irqsave(rq, &rf); update_rq_clock(rq); - update_cfs_rq_load_avg(cfs_rq_clock_pelt(cfs_rq), cfs_rq); + /* + * update_cfs_rq_load_avg() can call cpufreq_update_util(). Make sure + * that RT, DL and IRQ signals have been updated before updating CFS. + */ curr_class = rq->curr->sched_class; update_rt_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &rt_sched_class); update_dl_rq_load_avg(rq_clock_pelt(rq), rq, curr_class == &dl_sched_class); update_irq_load_avg(rq, 0); + + update_cfs_rq_load_avg(cfs_rq_clock_pelt(cfs_rq), cfs_rq); + update_blocked_load_status(rq, cfs_rq_has_blocked(cfs_rq) || others_have_blocked(rq)); rq_unlock_irqrestore(rq, &rf); }
| |