Skip to contentSkip to conversation

sched/isolation: Isolate unbound kthreads

PATCH Applied Frederic Weisbecker 3 participants 6 messages May 27, 2020 – Jun 16, 2020 Scheduler

lore
Share & tools
lore
#0 lore raw

cover sched/isolation: Isolate unbound kthreads

Kthreads are harder to affine and isolate than user tasks. They can't be placed inside cgroups/cpusets and the affinity for any newly created kthread is always overriden from the inherited kthreadd's affinity to system wide. Take that into account for nohz_full.

git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
sched/core

HEAD: 774258ad06e832e9fbdcf681225d2c29421a100f

Thanks,
Frederic
---

Marcelo Tosatti (2):
kthread: Switch to cpu_possible_mask
isolcpus: Affine unbound kernel threads to housekeeping cpus

 include/linux/sched/isolation.h | 1 +
 kernel/kthread.c                | 6 ++++--
 kernel/sched/isolation.c        | 3 ++-
 3 files changed, 7 insertions(+), 3 deletions(-)
#1 lore raw

PATCH 1/2 kthread: Switch to cpu_possible_mask

From: Marcelo Tosatti <[email protected]>

Next patch will switch unbound kernel threads mask to housekeeping_cpumask(), a subset of cpu_possible_mask. So in order to ease bisection, lets first switch kthreads default affinity from cpu_all_mask to cpu_possible_mask.

It looks safe to do so as cpu_possible_mask seem to be initialized at setup_arch() time, way before kthreadd is created.

Cc Chris Friesen <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Jim Somerville <[email protected]>
Cc Christoph Lameter <[email protected]>
---
 kernel/kthread.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/kthread.c +2 -2
index bfbfa481be3a..788bec94aca7 100644
@@ -360,7 +360,7 @@ struct task_struct *__kthread_create_on_node(int (*threadfn)(void *data),
 		 * The kernel thread should not inherit these properties.
 		 */
 		sched_setscheduler_nocheck(task, SCHED_NORMAL, &param);
-		set_cpus_allowed_ptr(task, cpu_all_mask);
+		set_cpus_allowed_ptr(task, cpu_possible_mask);
 	}
 	kfree(create);
 	return task;
@@ -585,7 +585,7 @@ int kthreadd(void *unused)
 	/* Setup a clean context for our children to inherit. */
 	set_task_comm(tsk, "kthreadd");
 	ignore_signals(tsk);
-	set_cpus_allowed_ptr(tsk, cpu_all_mask);
+	set_cpus_allowed_ptr(tsk, cpu_possible_mask);
 	set_mems_allowed(node_states[N_MEMORY]);
 
 	current->flags |= PF_NOFREEZE;
Signature · 1 line
2.26.2
#2 lore raw

[tip: sched/core] kthread: Switch to cpu_possible_mask

The following commit has been merged into the sched/core branch of tip:

The following commit has been merged into the sched/core branch of tip:

Commit-ID:     043eb8e1051143a24811e6f35c276e35ae8247b6
Gitweb:        https://git.kernel.org/tip/043eb8e1051143a24811e6f35c276e35ae8247b6
Author:        Marcelo Tosatti <[email protected]>
AuthorDate:    Wed, 27 May 2020 16:29:08 +02:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Mon, 15 Jun 2020 14:10:03 +02:00

kthread: Switch to cpu_possible_mask

Next patch will switch unbound kernel threads mask to housekeeping_cpumask(), a subset of cpu_possible_mask. So in order to ease bisection, lets first switch kthreads default affinity from cpu_all_mask to cpu_possible_mask.

It looks safe to do so as cpu_possible_mask seem to be initialized at setup_arch() time, way before kthreadd is created.

---
 kernel/kthread.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/kthread.c +2 -2
index 8e3d2d7..b86d37c 100644
@@ -383,7 +383,7 @@ struct task_struct *__kthread_create_on_node(int (*threadfn)(void *data),
 		 * The kernel thread should not inherit these properties.
 		 */
 		sched_setscheduler_nocheck(task, SCHED_NORMAL, &param);
-		set_cpus_allowed_ptr(task, cpu_all_mask);
+		set_cpus_allowed_ptr(task, cpu_possible_mask);
 	}
 	kfree(create);
 	return task;
@@ -608,7 +608,7 @@ int kthreadd(void *unused)
 	/* Setup a clean context for our children to inherit. */
 	set_task_comm(tsk, "kthreadd");
 	ignore_signals(tsk);
-	set_cpus_allowed_ptr(tsk, cpu_all_mask);
+	set_cpus_allowed_ptr(tsk, cpu_possible_mask);
 	set_mems_allowed(node_states[N_MEMORY]);
 
 	current->flags |= PF_NOFREEZE;
#3 lore raw

PATCH 2/2 isolcpus: Affine unbound kernel threads to housekeeping cpus

In reply to Frederic Weisbecker

From: Marcelo Tosatti <[email protected]>

This is a kernel enhancement that configures the cpu affinity of kernel threads via kernel boot option nohz_full=.

When this option is specified, the cpumask is immediately applied upon kthread launch. This does not affect kernel threads that specify cpu and node.

This allows CPU isolation (that is not allowing certain threads to execute on certain CPUs) without using the isolcpus=domain parameter, making it possible to enable load balancing on such CPUs during runtime (see kernel-parameters.txt).

Note-1: this is based off on Wind River's patch at
https://github.com/starlingx-staging/stx-integ/blob/master/kernel/kernel-std/centos/patches/affine-compute-kernel-threads.patch

Difference being that this patch is limited to modifying kernel thread cpumask. Behaviour of other threads can be controlled via cgroups or sched_setaffinity.

Note-2: Wind River's patch was based off Christoph Lameter's patch at
https://lwn.net/Articles/565932/ with the only difference being the kernel parameter changed from kthread to kthread_cpus.

Cc Chris Friesen <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Jim Somerville <[email protected]>
Cc Christoph Lameter <[email protected]>
---
 include/linux/sched/isolation.h | 1 +
 kernel/kthread.c                | 6 ++++--
 kernel/sched/isolation.c        | 3 ++-
 3 files changed, 7 insertions(+), 3 deletions(-)
include/linux/sched/isolation.h +1 -0
index 0fbcbacd1b29..cc9f393e2a70 100644
@@ -14,6 +14,7 @@ enum hk_flags {
 	HK_FLAG_DOMAIN		= (1 << 5),
 	HK_FLAG_WQ		= (1 << 6),
 	HK_FLAG_MANAGED_IRQ	= (1 << 7),
+	HK_FLAG_KTHREAD		= (1 << 8),
 };
 
 #ifdef CONFIG_CPU_ISOLATION
kernel/kthread.c +4 -2
index 788bec94aca7..eaeadefc0f8f 100644
@@ -23,6 +23,7 @@
 #include <linux/ptrace.h>
 #include <linux/uaccess.h>
 #include <linux/numa.h>
+#include <linux/sched/isolation.h>
 #include <trace/events/sched.h>
 
 static DEFINE_SPINLOCK(kthread_create_lock);
@@ -360,7 +361,8 @@ struct task_struct *__kthread_create_on_node(int (*threadfn)(void *data),
 		 * The kernel thread should not inherit these properties.
 		 */
 		sched_setscheduler_nocheck(task, SCHED_NORMAL, &param);
-		set_cpus_allowed_ptr(task, cpu_possible_mask);
+		set_cpus_allowed_ptr(task,
+				     housekeeping_cpumask(HK_FLAG_KTHREAD));
 	}
 	kfree(create);
 	return task;
@@ -585,7 +587,7 @@ int kthreadd(void *unused)
 	/* Setup a clean context for our children to inherit. */
 	set_task_comm(tsk, "kthreadd");
 	ignore_signals(tsk);
-	set_cpus_allowed_ptr(tsk, cpu_possible_mask);
+	set_cpus_allowed_ptr(tsk, housekeeping_cpumask(HK_FLAG_KTHREAD));
 	set_mems_allowed(node_states[N_MEMORY]);
 
 	current->flags |= PF_NOFREEZE;
kernel/sched/isolation.c +2 -1
index 808244f3ddd9..5a6ea03f9882 100644
@@ -140,7 +140,8 @@ static int __init housekeeping_nohz_full_setup(char *str)
 {
 	unsigned int flags;
 
-	flags = HK_FLAG_TICK | HK_FLAG_WQ | HK_FLAG_TIMER | HK_FLAG_RCU | HK_FLAG_MISC;
+	flags = HK_FLAG_TICK | HK_FLAG_WQ | HK_FLAG_TIMER | HK_FLAG_RCU |
+		HK_FLAG_MISC | HK_FLAG_KTHREAD;
 
 	return housekeeping_setup(str, flags);
 }
Signature · 1 line
2.26.2
#4 lore raw

[tip: sched/core] isolcpus: Affine unbound kernel threads to housekeeping cpus

The following commit has been merged into the sched/core branch of tip:

The following commit has been merged into the sched/core branch of tip:

Commit-ID:     9cc5b8656892a72438ee7deb5e80f5be47643b8b
Gitweb:        https://git.kernel.org/tip/9cc5b8656892a72438ee7deb5e80f5be47643b8b
Author:        Marcelo Tosatti <[email protected]>
AuthorDate:    Wed, 27 May 2020 16:29:09 +02:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Mon, 15 Jun 2020 14:10:03 +02:00

isolcpus: Affine unbound kernel threads to housekeeping cpus

This is a kernel enhancement that configures the cpu affinity of kernel threads via kernel boot option nohz_full=.

When this option is specified, the cpumask is immediately applied upon kthread launch. This does not affect kernel threads that specify cpu and node.

This allows CPU isolation (that is not allowing certain threads to execute on certain CPUs) without using the isolcpus=domain parameter, making it possible to enable load balancing on such CPUs during runtime (see kernel-parameters.txt).

Note-1: this is based off on Wind River's patch at
https://github.com/starlingx-staging/stx-integ/blob/master/kernel/kernel-std/centos/patches/affine-compute-kernel-threads.patch

Difference being that this patch is limited to modifying kernel thread cpumask. Behaviour of other threads can be controlled via cgroups or sched_setaffinity.

Note-2: Wind River's patch was based off Christoph Lameter's patch at
https://lwn.net/Articles/565932/ with the only difference being the kernel parameter changed from kthread to kthread_cpus.

---
 include/linux/sched/isolation.h | 1 +
 kernel/kthread.c                | 6 ++++--
 kernel/sched/isolation.c        | 3 ++-
 3 files changed, 7 insertions(+), 3 deletions(-)
include/linux/sched/isolation.h +1 -0
index 0fbcbac..cc9f393 100644
@@ -14,6 +14,7 @@ enum hk_flags {
 	HK_FLAG_DOMAIN		= (1 << 5),
 	HK_FLAG_WQ		= (1 << 6),
 	HK_FLAG_MANAGED_IRQ	= (1 << 7),
+	HK_FLAG_KTHREAD		= (1 << 8),
 };
 
 #ifdef CONFIG_CPU_ISOLATION
kernel/kthread.c +4 -2
index b86d37c..032b610 100644
@@ -27,6 +27,7 @@
 #include <linux/ptrace.h>
 #include <linux/uaccess.h>
 #include <linux/numa.h>
+#include <linux/sched/isolation.h>
 #include <trace/events/sched.h>
 
 
@@ -383,7 +384,8 @@ struct task_struct *__kthread_create_on_node(int (*threadfn)(void *data),
 		 * The kernel thread should not inherit these properties.
 		 */
 		sched_setscheduler_nocheck(task, SCHED_NORMAL, &param);
-		set_cpus_allowed_ptr(task, cpu_possible_mask);
+		set_cpus_allowed_ptr(task,
+				     housekeeping_cpumask(HK_FLAG_KTHREAD));
 	}
 	kfree(create);
 	return task;
@@ -608,7 +610,7 @@ int kthreadd(void *unused)
 	/* Setup a clean context for our children to inherit. */
 	set_task_comm(tsk, "kthreadd");
 	ignore_signals(tsk);
-	set_cpus_allowed_ptr(tsk, cpu_possible_mask);
+	set_cpus_allowed_ptr(tsk, housekeeping_cpumask(HK_FLAG_KTHREAD));
 	set_mems_allowed(node_states[N_MEMORY]);
 
 	current->flags |= PF_NOFREEZE;
kernel/sched/isolation.c +2 -1
index 808244f..5a6ea03 100644
@@ -140,7 +140,8 @@ static int __init housekeeping_nohz_full_setup(char *str)
 {
 	unsigned int flags;
 
-	flags = HK_FLAG_TICK | HK_FLAG_WQ | HK_FLAG_TIMER | HK_FLAG_RCU | HK_FLAG_MISC;
+	flags = HK_FLAG_TICK | HK_FLAG_WQ | HK_FLAG_TIMER | HK_FLAG_RCU |
+		HK_FLAG_MISC | HK_FLAG_KTHREAD;
 
 	return housekeeping_setup(str, flags);
 }
maintainer
#5 lore raw
In reply to Frederic Weisbecker
Quoted 5 lines · Frederic Weisbecker

Kthreads are harder to affine and isolate than user tasks. They can't be placed inside cgroups/cpusets and the affinity for any newly created kthread is always overriden from the inherited kthreadd's affinity to system wide. Take that into account for nohz_full.

Thanks!