Update NEST for 5.15, add THP Shrinker and ksm for 6.0

Signed-off-by: Peter Jung <[email protected]>

5.15/0001-NEST.patch

@@ -1,6 +1,6 @@
-From 89923bf3230fb3ed867afc0a442eef69252044e9 Mon Sep 17 00:00:00 2001
+From e87fabf12ab7b61d05718076efd69a49433da2b4 Mon Sep 17 00:00:00 2001
 From: Peter Jung <[email protected]>
-Date: Mon, 19 Sep 2022 18:16:16 +0200
+Date: Wed, 12 Oct 2022 00:05:59 +0200
 Subject: [PATCH] NEST
 
 Signed-off-by: Peter Jung <[email protected]>
@@ -9,10 +9,10 @@ Signed-off-by: Peter Jung <[email protected]>
  include/linux/sched.h          |  14 ++
  include/linux/sched/topology.h |   2 +
  kernel/sched/core.c            |  86 +++++++++-
- kernel/sched/fair.c            | 297 ++++++++++++++++++++++++++++++++-
+ kernel/sched/fair.c            | 299 ++++++++++++++++++++++++++++++++-
  kernel/sched/idle.c            |  20 +++
  kernel/sched/sched.h           |  49 ++++++
- 7 files changed, 455 insertions(+), 15 deletions(-)
+ 7 files changed, 457 insertions(+), 15 deletions(-)
 
 diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
 index 85f6e242b6b4..8520b84d2046 100644
@@ -73,7 +73,7 @@ index 63a04a65e310..79bf1f6c7b23 100644
 
  struct sched_domain {
 diff --git a/kernel/sched/core.c b/kernel/sched/core.c
-index 85be684687b0..0f0ebebafe3b 100644
+index 85be684687b0..fb7840a27cf7 100644
 --- a/kernel/sched/core.c
 +++ b/kernel/sched/core.c
 @@ -2495,6 +2495,7 @@ __do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask, u32
@@ -91,7 +91,7 @@ index 85be684687b0..0f0ebebafe3b 100644
 -	if (unlikely(!is_cpu_allowed(p, cpu)))
 +	if (unlikely(!is_cpu_allowed(p, cpu))) {
 +		if (p->use_expand_mask)
-+			atomic_set(&cpu_rq(cpu)->taken,0);
++			atomic_set(&cpu_rq(p->thread_info.cpu)->taken,0);
  		cpu = select_fallback_rq(task_cpu(p), p);
 +	}
 
@@ -156,7 +156,7 @@ index 85be684687b0..0f0ebebafe3b 100644
  	post_init_entity_util_avg(p);
 
 +	if (p->use_expand_mask)
-+		atomic_set(&cpu_rq(p->cpu)->taken,0);
++		atomic_set(&cpu_rq(p->thread_info.cpu)->taken,0);
  	activate_task(rq, p, ENQUEUE_NOCLOCK);
  	trace_sched_wakeup_new(p);
  	check_preempt_curr(rq, p, WF_FORK);
@@ -269,7 +269,7 @@ index 85be684687b0..0f0ebebafe3b 100644
 
  		rq_attach_root(rq, &def_root_domain);
 diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
-index a853e4e9e3c3..0a61e53fc874 100644
+index a853e4e9e3c3..9857c18956b6 100644
 --- a/kernel/sched/fair.c
 +++ b/kernel/sched/fair.c
 @@ -3378,6 +3378,61 @@ static inline void update_tg_load_avg(struct cfs_rq *cfs_rq)
@@ -446,19 +446,20 @@ index a853e4e9e3c3..0a61e53fc874 100644
 
  	/*
  	 * On asymmetric system, update task utilization because we will check
-@@ -6513,7 +6619,39 @@ static int select_idle_sibling(struct task_struct *p, int prev, int target)
+@@ -6513,7 +6619,41 @@ static int select_idle_sibling(struct task_struct *p, int prev, int target)
  	if ((unsigned)i < nr_cpumask_bits)
  		return i;
 
 -	return target;
++	if (!sd->parent) goto abort;
 +	group0 = group = sd->parent->groups;
 +	do {
 +		struct cpumask *m = sched_group_span(group);
 +
 +		if (cpumask_test_cpu(otarget, m))
 +			goto next;
 +
-+		sd1 = rcu_dereference(per_cpu(sd_llc, group_first_cpu(group)));
++		sd1 = rcu_dereference(per_cpu(sd_llc, cpumask_first(sched_group_span(group))));
 +		if (!sd1 || !READ_ONCE(sd1->shared->has_idle_threads))
 +			goto next;
 +
@@ -483,19 +484,20 @@ index a853e4e9e3c3..0a61e53fc874 100644
 +	} while (group != group0);
 +
 +
++abort:
 +	return otarget;
  }
 
  /**
-@@ -6940,6 +7078,7 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
+@@ -6940,6 +7080,7 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
  	int cpu = smp_processor_id();
  	int new_cpu = prev_cpu;
  	int want_affine = 0;
 +	int impatient = 0;
  	/* SD_flags and WF_flags share the first nibble */
  	int sd_flag = wake_flags & 0xF;
 
-@@ -6961,6 +7100,144 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
+@@ -6961,6 +7102,144 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
  	}
 
  	rcu_read_lock();
@@ -640,7 +642,7 @@ index a853e4e9e3c3..0a61e53fc874 100644
  	for_each_domain(cpu, tmp) {
  		/*
  		 * If both 'cpu' and 'prev_cpu' are part of this domain,
-@@ -6988,6 +7265,14 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
+@@ -6988,6 +7267,14 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int wake_flags)
  		/* Fast path */
  		new_cpu = select_idle_sibling(p, prev_cpu, new_cpu);
  	}
@@ -656,7 +658,7 @@ index a853e4e9e3c3..0a61e53fc874 100644
 
  	return new_cpu;
 diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
-index d17b0a5ce6ac..22dda7a58481 100644
+index 499a3e286cd0..cf6a3ceb4c79 100644
 --- a/kernel/sched/idle.c
 +++ b/kernel/sched/idle.c
 @@ -262,6 +262,26 @@ static void do_idle(void)
@@ -754,5 +756,5 @@ index e49902898253..ee9a744377c0 100644
  static inline struct task_struct *task_of(struct sched_entity *se)
  {
 -- 
-2.37.3
+2.38.0