Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1632897 > unrolled thread

[PATCH 00/14] sched/topology fixes

Started byPeter Zijlstra <peterz@infradead.org>
First post2017-04-28 15:40 +0200
Last post2017-05-02 17:00 +0200
Articles 6 — 2 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH 00/14] sched/topology fixes Peter Zijlstra <peterz@infradead.org> - 2017-04-28 15:40 +0200
    Re: [PATCH 00/14] sched/topology fixes Peter Zijlstra <peterz@infradead.org> - 2017-04-28 16:00 +0200
      Re: [PATCH 00/14] sched/topology fixes Lauro Venancio <lvenanci@redhat.com> - 2017-04-29 01:40 +0200
        Re: [PATCH 00/14] sched/topology fixes Peter Zijlstra <peterz@infradead.org> - 2017-05-01 11:00 +0200
      Re: [PATCH 00/14] sched/topology fixes Peter Zijlstra <peterz@infradead.org> - 2017-05-02 16:50 +0200
        Re: [PATCH 00/14] sched/topology fixes Lauro Venancio <lvenanci@redhat.com> - 2017-05-02 17:00 +0200

#1632897 — [PATCH 00/14] sched/topology fixes

FromPeter Zijlstra <peterz@infradead.org>
Date2017-04-28 15:40 +0200
Subject[PATCH 00/14] sched/topology fixes
Message-ID<tBehP-7VW-5@gated-at.bofh.it>
Hi,

These patches are based upon the hard work of Lauro. He put in the time and
effort to understand and debug the code.

So while I didn't take many of his actual patches; I want to thank him for
doing the work. Hopefully the "Debugged-by:" tag conveys some of that.

In any case, please have a look. I think these should about cover things.

Rik, Lauro, could you guys in particular look at the final patch that adds a
few comments. I attempted to document the intent and understanding there. But
given I've been staring at this stuff too long I could've missed the obvious.

Comments and or suggestions welcome.

[toc] | [next] | [standalone]


#1632906

FromPeter Zijlstra <peterz@infradead.org>
Date2017-04-28 16:00 +0200
Message-ID<tBeBb-84B-1@gated-at.bofh.it>
In reply to#1632897
On Fri, Apr 28, 2017 at 03:19:58PM +0200, Peter Zijlstra wrote:
> Hi,
> 
> These patches are based upon the hard work of Lauro. He put in the time and
> effort to understand and debug the code.
> 
> So while I didn't take many of his actual patches; I want to thank him for
> doing the work. Hopefully the "Debugged-by:" tag conveys some of that.
> 
> In any case, please have a look. I think these should about cover things.
> 
> Rik, Lauro, could you guys in particular look at the final patch that adds a
> few comments. I attempted to document the intent and understanding there. But
> given I've been staring at this stuff too long I could've missed the obvious.
> 
> Comments and or suggestions welcome.
> 

Also, the following occurred to me:

  sg_span & sg_mask == sg_mask

Therefore, we don't need to do the whole "sg_span &" business.

Hmm?

---
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7996,7 +7996,7 @@ static int active_load_balance_cpu_stop(
 static int should_we_balance(struct lb_env *env)
 {
 	struct sched_group *sg = env->sd->groups;
-	struct cpumask *sg_cpus, *sg_mask;
+	struct cpumask *sg_mask;
 	int cpu, balance_cpu = -1;
 
 	/*
@@ -8006,11 +8006,10 @@ static int should_we_balance(struct lb_e
 	if (env->idle == CPU_NEWLY_IDLE)
 		return 1;
 
-	sg_cpus = sched_group_cpus(sg);
 	sg_mask = sched_group_mask(sg);
 	/* Try to find first idle cpu */
-	for_each_cpu_and(cpu, sg_cpus, env->cpus) {
-		if (!cpumask_test_cpu(cpu, sg_mask) || !idle_cpu(cpu))
+	for_each_cpu_and(cpu, sg_mask, env->cpus) {
+		if (!idle_cpu(cpu))
 			continue;
 
 		balance_cpu = cpu;
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -85,7 +85,8 @@ static int sched_domain_debug_one(struct
 				group->sgc->id,
 				cpumask_pr_args(sched_group_cpus(group)));
 
-		if ((sd->flags & SD_OVERLAP) && !cpumask_full(sched_group_mask(group))) {
+		if ((sd->flags & SD_OVERLAP) &&
+		    !cpumask_equal(sched_group_mask(group), sched_group_cpus(group))) {
 			printk(KERN_CONT " mask=%*pbl",
 				cpumask_pr_args(sched_group_mask(group)));
 		}
@@ -505,7 +506,7 @@ enum s_alloc {
  */
 int group_balance_cpu(struct sched_group *sg)
 {
-	return cpumask_first_and(sched_group_cpus(sg), sched_group_mask(sg));
+	return cpumask_first(sched_group_mask(sg));
 }
 
 
@@ -856,7 +857,7 @@ build_sched_groups(struct sched_domain *
 			continue;
 
 		group = get_group(i, sdd, &sg);
-		cpumask_setall(sched_group_mask(sg));
+		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
 
 		for_each_cpu(j, span) {
 			if (get_group(j, sdd, NULL) != group)

[toc] | [prev] | [next] | [standalone]


#1633204

FromLauro Venancio <lvenanci@redhat.com>
Date2017-04-29 01:40 +0200
Message-ID<tBnEt-6fw-3@gated-at.bofh.it>
In reply to#1632906
On 04/28/2017 10:53 AM, Peter Zijlstra wrote:
> On Fri, Apr 28, 2017 at 03:19:58PM +0200, Peter Zijlstra wrote:
>> Hi,
>>
>> These patches are based upon the hard work of Lauro. He put in the time and
>> effort to understand and debug the code.
>>
>> So while I didn't take many of his actual patches; I want to thank him for
>> doing the work. Hopefully the "Debugged-by:" tag conveys some of that.
>>
>> In any case, please have a look. I think these should about cover things.
>>
>> Rik, Lauro, could you guys in particular look at the final patch that adds a
>> few comments. I attempted to document the intent and understanding there. But
>> given I've been staring at this stuff too long I could've missed the obvious.
>>
>> Comments and or suggestions welcome.
>>
> Also, the following occurred to me:
>
>   sg_span & sg_mask == sg_mask
>
> Therefore, we don't need to do the whole "sg_span &" business.
>
> Hmm?
Agreed. Maybe we have to rename "mask" to something else. Maybe
"group_reached_by_cpus" or "group_installed_on_cpus".

>
> ---
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -7996,7 +7996,7 @@ static int active_load_balance_cpu_stop(
>  static int should_we_balance(struct lb_env *env)
>  {
>  	struct sched_group *sg = env->sd->groups;
> -	struct cpumask *sg_cpus, *sg_mask;
> +	struct cpumask *sg_mask;
>  	int cpu, balance_cpu = -1;
>  
>  	/*
> @@ -8006,11 +8006,10 @@ static int should_we_balance(struct lb_e
>  	if (env->idle == CPU_NEWLY_IDLE)
>  		return 1;
>  
> -	sg_cpus = sched_group_cpus(sg);
>  	sg_mask = sched_group_mask(sg);
>  	/* Try to find first idle cpu */
> -	for_each_cpu_and(cpu, sg_cpus, env->cpus) {
> -		if (!cpumask_test_cpu(cpu, sg_mask) || !idle_cpu(cpu))
> +	for_each_cpu_and(cpu, sg_mask, env->cpus) {
> +		if (!idle_cpu(cpu))
>  			continue;
>  
>  		balance_cpu = cpu;
> --- a/kernel/sched/topology.c
> +++ b/kernel/sched/topology.c
> @@ -85,7 +85,8 @@ static int sched_domain_debug_one(struct
>  				group->sgc->id,
>  				cpumask_pr_args(sched_group_cpus(group)));
>  
> -		if ((sd->flags & SD_OVERLAP) && !cpumask_full(sched_group_mask(group))) {
> +		if ((sd->flags & SD_OVERLAP) &&
> +		    !cpumask_equal(sched_group_mask(group), sched_group_cpus(group))) {
>  			printk(KERN_CONT " mask=%*pbl",
>  				cpumask_pr_args(sched_group_mask(group)));
>  		}
> @@ -505,7 +506,7 @@ enum s_alloc {
>   */
>  int group_balance_cpu(struct sched_group *sg)
>  {
> -	return cpumask_first_and(sched_group_cpus(sg), sched_group_mask(sg));
> +	return cpumask_first(sched_group_mask(sg));
>  }
>  
>  
> @@ -856,7 +857,7 @@ build_sched_groups(struct sched_domain *
>  			continue;
>  
>  		group = get_group(i, sdd, &sg);
> -		cpumask_setall(sched_group_mask(sg));
> +		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
>  
>  		for_each_cpu(j, span) {
>  			if (get_group(j, sdd, NULL) != group)

[toc] | [prev] | [next] | [standalone]


#1633579

FromPeter Zijlstra <peterz@infradead.org>
Date2017-05-01 11:00 +0200
Message-ID<tCflv-6TW-1@gated-at.bofh.it>
In reply to#1633204
On Fri, Apr 28, 2017 at 08:30:05PM -0300, Lauro Venancio wrote:
> On 04/28/2017 10:53 AM, Peter Zijlstra wrote:
> > Also, the following occurred to me:
> >
> >   sg_span & sg_mask == sg_mask
> >
> > Therefore, we don't need to do the whole "sg_span &" business.
> >
> > Hmm?

> Agreed. Maybe we have to rename "mask" to something else. Maybe
> "group_reached_by_cpus" or "group_installed_on_cpus".

I went with group_balance_mask() to match the existing
group_balance_cpu().

And now the temptation is very great to also fix up the
sched_group_cpus() vs sched_domain_span() thing...

---
Subject: sched/topology: Rename sched_group_mask()
From: Peter Zijlstra <peterz@infradead.org>
Date: Mon May 1 10:47:02 CEST 2017

Since sched_group_mask() is now an independent cpumask (it no longer
masks sched_group_cpus()), rename the thing.

Suggested-by: Lauro Ramos Venancio <lvenanci@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 kernel/sched/fair.c     |    4 +---
 kernel/sched/sched.h    |    7 +++----
 kernel/sched/topology.c |   31 ++++++++++++++-----------------
 3 files changed, 18 insertions(+), 24 deletions(-)

--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7996,7 +7996,6 @@ static int active_load_balance_cpu_stop(
 static int should_we_balance(struct lb_env *env)
 {
 	struct sched_group *sg = env->sd->groups;
-	struct cpumask *sg_mask;
 	int cpu, balance_cpu = -1;
 
 	/*
@@ -8006,9 +8005,8 @@ static int should_we_balance(struct lb_e
 	if (env->idle == CPU_NEWLY_IDLE)
 		return 1;
 
-	sg_mask = sched_group_mask(sg);
 	/* Try to find first idle cpu */
-	for_each_cpu_and(cpu, sg_mask, env->cpus) {
+	for_each_cpu_and(cpu, group_balance_mask(sg), env->cpus) {
 		if (!idle_cpu(cpu))
 			continue;
 
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1027,7 +1027,7 @@ struct sched_group_capacity {
 	int id;
 #endif
 
-	unsigned long cpumask[0]; /* iteration mask */
+	unsigned long cpumask[0]; /* balance mask */
 };
 
 struct sched_group {
@@ -1054,10 +1054,9 @@ static inline struct cpumask *sched_grou
 }
 
 /*
- * cpumask masking which cpus in the group are allowed to iterate up the domain
- * tree.
+ * See build_balance_mask().
  */
-static inline struct cpumask *sched_group_mask(struct sched_group *sg)
+static inline struct cpumask *group_balance_mask(struct sched_group *sg)
 {
 	return to_cpumask(sg->sgc->cpumask);
 }
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -86,9 +86,9 @@ static int sched_domain_debug_one(struct
 				cpumask_pr_args(sched_group_cpus(group)));
 
 		if ((sd->flags & SD_OVERLAP) &&
-		    !cpumask_equal(sched_group_mask(group), sched_group_cpus(group))) {
+		    !cpumask_equal(group_balance_mask(group), sched_group_cpus(group))) {
 			printk(KERN_CONT " mask=%*pbl",
-				cpumask_pr_args(sched_group_mask(group)));
+				cpumask_pr_args(group_balance_mask(group)));
 		}
 
 		if (group->sgc->capacity != SCHED_CAPACITY_SCALE)
@@ -497,16 +497,16 @@ enum s_alloc {
 
 /*
  * Return the canonical balance CPU for this group, this is the first CPU
- * of this group that's also in the iteration mask.
+ * of this group that's also in the balance mask.
  *
- * The iteration mask are all those CPUs that could actually end up at this
- * group. See build_group_mask().
+ * The balance mask are all those CPUs that could actually end up at this
+ * group. See build_balance_mask().
  *
  * Also see should_we_balance().
  */
 int group_balance_cpu(struct sched_group *sg)
 {
-	return cpumask_first(sched_group_mask(sg));
+	return cpumask_first(group_balance_mask(sg));
 }
 
 
@@ -563,7 +563,7 @@ int group_balance_cpu(struct sched_group
  * groups include the CPUs of Node-0, while those CPUs would not in fact ever
  * end up at those groups (they would end up in group: 0-1,3).
  *
- * To correct this we have to introduce the group iteration mask. This mask
+ * To correct this we have to introduce the group balance mask. This mask
  * will contain those CPUs in the group that can reach this group given the
  * (child) domain tree.
  *
@@ -607,11 +607,8 @@ int group_balance_cpu(struct sched_group
 
 
 /*
- * Build an iteration mask that can exclude certain CPUs from the upwards
- * domain traversal.
- *
- * Only CPUs that can arrive at this group should be considered to continue
- * balancing.
+ * Build the balance mask; it contains only those CPUs that can arrive at this
+ * group and should be considered to continue balancing.
  *
  * We do this during the group creation pass, therefore the group information
  * isn't complete yet, however since each group represents a (child) domain we
@@ -619,7 +616,7 @@ int group_balance_cpu(struct sched_group
  * complete).
  */
 static void
-build_group_mask(struct sched_domain *sd, struct sched_group *sg, struct cpumask *mask)
+build_balance_mask(struct sched_domain *sd, struct sched_group *sg, struct cpumask *mask)
 {
 	const struct cpumask *sg_span = sched_group_cpus(sg);
 	struct sd_data *sdd = sd->private;
@@ -684,14 +681,14 @@ static void init_overlap_sched_group(str
 	struct cpumask *sg_span;
 	int cpu;
 
-	build_group_mask(sd, sg, mask);
+	build_balance_mask(sd, sg, mask);
 	cpu = cpumask_first_and(sched_group_cpus(sg), mask);
 
 	sg->sgc = *per_cpu_ptr(sdd->sgc, cpu);
 	if (atomic_inc_return(&sg->sgc->ref) == 1)
-		cpumask_copy(sched_group_mask(sg), mask);
+		cpumask_copy(group_balance_mask(sg), mask);
 	else
-		WARN_ON_ONCE(!cpumask_equal(sched_group_mask(sg), mask));
+		WARN_ON_ONCE(!cpumask_equal(group_balance_mask(sg), mask));
 
 	/*
 	 * Initialize sgc->capacity such that even if we mess up the
@@ -888,7 +885,7 @@ build_sched_groups(struct sched_domain *
 			continue;
 
 		group = get_group(i, sdd, &sg);
-		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
+		cpumask_copy(group_balance_mask(sg), sched_group_cpus(sg));
 
 		for_each_cpu(j, span) {
 			if (get_group(j, sdd, NULL) != group)

[toc] | [prev] | [next] | [standalone]


#1634463

FromPeter Zijlstra <peterz@infradead.org>
Date2017-05-02 16:50 +0200
Message-ID<tCHhL-7T4-5@gated-at.bofh.it>
In reply to#1632906
On Fri, Apr 28, 2017 at 03:53:39PM +0200, Peter Zijlstra wrote:
> Also, the following occurred to me:
> 
>   sg_span & sg_mask == sg_mask
> 
> Therefore, we don't need to do the whole "sg_span &" business.
> 
> Hmm?

> @@ -856,7 +857,7 @@ build_sched_groups(struct sched_domain *
>  			continue;
>  
>  		group = get_group(i, sdd, &sg);
> -		cpumask_setall(sched_group_mask(sg));
> +		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
>  
>  		for_each_cpu(j, span) {
>  			if (get_group(j, sdd, NULL) != group)

OK, so this explodes mightily.

That code also hurt my brain bad, so I had to fix that a little.

The below seems to boot.

---
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7996,7 +7996,7 @@ static int active_load_balance_cpu_stop(
 static int should_we_balance(struct lb_env *env)
 {
 	struct sched_group *sg = env->sd->groups;
-	struct cpumask *sg_cpus, *sg_mask;
+	struct cpumask *sg_mask;
 	int cpu, balance_cpu = -1;
 
 	/*
@@ -8006,11 +8006,10 @@ static int should_we_balance(struct lb_e
 	if (env->idle == CPU_NEWLY_IDLE)
 		return 1;
 
-	sg_cpus = sched_group_cpus(sg);
 	sg_mask = sched_group_mask(sg);
 	/* Try to find first idle cpu */
-	for_each_cpu_and(cpu, sg_cpus, env->cpus) {
-		if (!cpumask_test_cpu(cpu, sg_mask) || !idle_cpu(cpu))
+	for_each_cpu_and(cpu, sg_mask, env->cpus) {
+		if (!idle_cpu(cpu))
 			continue;
 
 		balance_cpu = cpu;
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -85,7 +85,8 @@ static int sched_domain_debug_one(struct
 				group->sgc->id,
 				cpumask_pr_args(sched_group_cpus(group)));
 
-		if ((sd->flags & SD_OVERLAP) && !cpumask_full(sched_group_mask(group))) {
+		if ((sd->flags & SD_OVERLAP) &&
+		    !cpumask_equal(sched_group_mask(group), sched_group_cpus(group))) {
 			printk(KERN_CONT " mask=%*pbl",
 				cpumask_pr_args(sched_group_mask(group)));
 		}
@@ -505,7 +506,7 @@ enum s_alloc {
  */
 int group_balance_cpu(struct sched_group *sg)
 {
-	return cpumask_first_and(sched_group_cpus(sg), sched_group_mask(sg));
+	return cpumask_first(sched_group_mask(sg));
 }
 
 
@@ -833,23 +834,34 @@ build_overlap_sched_groups(struct sched_
  * [*] in other words, the first group of each domain is its child domain.
  */
 
-static int get_group(int cpu, struct sd_data *sdd, struct sched_group **sg)
+static struct sched_group *get_group(int cpu, struct sd_data *sdd)
 {
 	struct sched_domain *sd = *per_cpu_ptr(sdd->sd, cpu);
 	struct sched_domain *child = sd->child;
+	struct sched_group *sg;
 
 	if (child)
 		cpu = cpumask_first(sched_domain_span(child));
 
-	if (sg) {
-		*sg = *per_cpu_ptr(sdd->sg, cpu);
-		(*sg)->sgc = *per_cpu_ptr(sdd->sgc, cpu);
+	sg = *per_cpu_ptr(sdd->sg, cpu);
+	sg->sgc = *per_cpu_ptr(sdd->sgc, cpu);
 
-		/* For claim_allocations: */
-		atomic_set(&(*sg)->sgc->ref, 1);
+	/* For claim_allocations: */
+	atomic_inc(&sg->ref);
+	atomic_inc(&sg->sgc->ref);
+
+	if (child) {
+		cpumask_copy(sched_group_cpus(sg), sched_domain_span(child));
+		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
+	} else {
+		cpumask_set_cpu(cpu, sched_group_cpus(sg));
+		cpumask_set_cpu(cpu, sched_group_cpus(sg));
 	}
 
-	return cpu;
+	sg->sgc->capacity = SCHED_CAPACITY_SCALE * cpumask_weight(sched_group_cpus(sg));
+	sg->sgc->min_capacity = SCHED_CAPACITY_SCALE;
+
+	return sg;
 }
 
 /*
@@ -868,34 +880,20 @@ build_sched_groups(struct sched_domain *
 	struct cpumask *covered;
 	int i;
 
-	get_group(cpu, sdd, &sd->groups);
-	atomic_inc(&sd->groups->ref);
-
-	if (cpu != cpumask_first(span))
-		return 0;
-
 	lockdep_assert_held(&sched_domains_mutex);
 	covered = sched_domains_tmpmask;
 
 	cpumask_clear(covered);
 
-	for_each_cpu(i, span) {
+	for_each_cpu_wrap(i, span, cpu) {
 		struct sched_group *sg;
-		int group, j;
 
 		if (cpumask_test_cpu(i, covered))
 			continue;
 
-		group = get_group(i, sdd, &sg);
-		cpumask_setall(sched_group_mask(sg));
+		sg = get_group(i, sdd);
 
-		for_each_cpu(j, span) {
-			if (get_group(j, sdd, NULL) != group)
-				continue;
-
-			cpumask_set_cpu(j, covered);
-			cpumask_set_cpu(j, sched_group_cpus(sg));
-		}
+		cpumask_or(covered, covered, sched_group_cpus(sg));
 
 		if (!first)
 			first = sg;
@@ -904,6 +902,7 @@ build_sched_groups(struct sched_domain *
 		last = sg;
 	}
 	last->next = first;
+	sd->groups = first;
 
 	return 0;
 }

[toc] | [prev] | [next] | [standalone]


#1634472

FromLauro Venancio <lvenanci@redhat.com>
Date2017-05-02 17:00 +0200
Message-ID<tCHrs-7Ws-21@gated-at.bofh.it>
In reply to#1634463
On 05/02/2017 11:43 AM, Peter Zijlstra wrote:
> On Fri, Apr 28, 2017 at 03:53:39PM +0200, Peter Zijlstra wrote:
>> Also, the following occurred to me:
>>
>>   sg_span & sg_mask == sg_mask
>>
>> Therefore, we don't need to do the whole "sg_span &" business.
>>
>> Hmm?
>> @@ -856,7 +857,7 @@ build_sched_groups(struct sched_domain *
>>  			continue;
>>  
>>  		group = get_group(i, sdd, &sg);
>> -		cpumask_setall(sched_group_mask(sg));
>> +		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
>>  
>>  		for_each_cpu(j, span) {
>>  			if (get_group(j, sdd, NULL) != group)
> OK, so this explodes mightily.
>
> That code also hurt my brain bad, so I had to fix that a little.
>
> The below seems to boot.
>
> ---
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -7996,7 +7996,7 @@ static int active_load_balance_cpu_stop(
>  static int should_we_balance(struct lb_env *env)
>  {
>  	struct sched_group *sg = env->sd->groups;
> -	struct cpumask *sg_cpus, *sg_mask;
> +	struct cpumask *sg_mask;
>  	int cpu, balance_cpu = -1;
>  
>  	/*
> @@ -8006,11 +8006,10 @@ static int should_we_balance(struct lb_e
>  	if (env->idle == CPU_NEWLY_IDLE)
>  		return 1;
>  
> -	sg_cpus = sched_group_cpus(sg);
>  	sg_mask = sched_group_mask(sg);
>  	/* Try to find first idle cpu */
> -	for_each_cpu_and(cpu, sg_cpus, env->cpus) {
> -		if (!cpumask_test_cpu(cpu, sg_mask) || !idle_cpu(cpu))
> +	for_each_cpu_and(cpu, sg_mask, env->cpus) {
> +		if (!idle_cpu(cpu))
>  			continue;
>  
>  		balance_cpu = cpu;
> --- a/kernel/sched/topology.c
> +++ b/kernel/sched/topology.c
> @@ -85,7 +85,8 @@ static int sched_domain_debug_one(struct
>  				group->sgc->id,
>  				cpumask_pr_args(sched_group_cpus(group)));
>  
> -		if ((sd->flags & SD_OVERLAP) && !cpumask_full(sched_group_mask(group))) {
> +		if ((sd->flags & SD_OVERLAP) &&
> +		    !cpumask_equal(sched_group_mask(group), sched_group_cpus(group))) {
>  			printk(KERN_CONT " mask=%*pbl",
>  				cpumask_pr_args(sched_group_mask(group)));
>  		}
> @@ -505,7 +506,7 @@ enum s_alloc {
>   */
>  int group_balance_cpu(struct sched_group *sg)
>  {
> -	return cpumask_first_and(sched_group_cpus(sg), sched_group_mask(sg));
> +	return cpumask_first(sched_group_mask(sg));
>  }
>  
>  
> @@ -833,23 +834,34 @@ build_overlap_sched_groups(struct sched_
>   * [*] in other words, the first group of each domain is its child domain.
>   */
>  
> -static int get_group(int cpu, struct sd_data *sdd, struct sched_group **sg)
> +static struct sched_group *get_group(int cpu, struct sd_data *sdd)
>  {
>  	struct sched_domain *sd = *per_cpu_ptr(sdd->sd, cpu);
>  	struct sched_domain *child = sd->child;
> +	struct sched_group *sg;
>  
>  	if (child)
>  		cpu = cpumask_first(sched_domain_span(child));
>  
> -	if (sg) {
> -		*sg = *per_cpu_ptr(sdd->sg, cpu);
> -		(*sg)->sgc = *per_cpu_ptr(sdd->sgc, cpu);
> +	sg = *per_cpu_ptr(sdd->sg, cpu);
> +	sg->sgc = *per_cpu_ptr(sdd->sgc, cpu);
>  
> -		/* For claim_allocations: */
> -		atomic_set(&(*sg)->sgc->ref, 1);
> +	/* For claim_allocations: */
> +	atomic_inc(&sg->ref);
> +	atomic_inc(&sg->sgc->ref);
> +
> +	if (child) {
> +		cpumask_copy(sched_group_cpus(sg), sched_domain_span(child));
> +		cpumask_copy(sched_group_mask(sg), sched_group_cpus(sg));
> +	} else {
> +		cpumask_set_cpu(cpu, sched_group_cpus(sg));
> +		cpumask_set_cpu(cpu, sched_group_cpus(sg));
Typo here. The mask is not being set in the else clause.

>  	}
>  
> -	return cpu;
> +	sg->sgc->capacity = SCHED_CAPACITY_SCALE * cpumask_weight(sched_group_cpus(sg));
> +	sg->sgc->min_capacity = SCHED_CAPACITY_SCALE;
> +
> +	return sg;
>  }
>  
>  /*
> @@ -868,34 +880,20 @@ build_sched_groups(struct sched_domain *
>  	struct cpumask *covered;
>  	int i;
>  
> -	get_group(cpu, sdd, &sd->groups);
> -	atomic_inc(&sd->groups->ref);
> -
> -	if (cpu != cpumask_first(span))
> -		return 0;
> -
>  	lockdep_assert_held(&sched_domains_mutex);
>  	covered = sched_domains_tmpmask;
>  
>  	cpumask_clear(covered);
>  
> -	for_each_cpu(i, span) {
> +	for_each_cpu_wrap(i, span, cpu) {
>  		struct sched_group *sg;
> -		int group, j;
>  
>  		if (cpumask_test_cpu(i, covered))
>  			continue;
>  
> -		group = get_group(i, sdd, &sg);
> -		cpumask_setall(sched_group_mask(sg));
> +		sg = get_group(i, sdd);
>  
> -		for_each_cpu(j, span) {
> -			if (get_group(j, sdd, NULL) != group)
> -				continue;
> -
> -			cpumask_set_cpu(j, covered);
> -			cpumask_set_cpu(j, sched_group_cpus(sg));
> -		}
> +		cpumask_or(covered, covered, sched_group_cpus(sg));
>  
>  		if (!first)
>  			first = sg;
> @@ -904,6 +902,7 @@ build_sched_groups(struct sched_domain *
>  		last = sg;
>  	}
>  	last->next = first;
> +	sd->groups = first;
>  
>  	return 0;
>  }

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web