[PATCH/RESEND 0/2] sched: re-add removed arch_update_cpu_topology call

[PATCH/RESEND 0/2] sched: re-add removed arch_update_cpu_topology call

[Heiko Carstens]

This fixes a regression introduced with
e761b7725234276a802322549cee5255305a0930 ("cpu hotplug, sched:
Introduce cpu_active_map and redo sched domain managment (take 2)":

arch_reinit_sched_domains used to call arch_update_cpu_topology
via arch_init_sched_domains. This call got lost with the above
mentioned patch.

So we might end up with outdated and missing cpus in the cpu core
maps (s390 used to call arch_reinit_sched_domains if cpu topology changed).

This patch set changes arch_update_cpu_topology so that it can indicate
if the cpu topology changed and adds a call to arch_update_cpu_topology
in partition_sched_domains which gets called whenever scheduling domains
get updated.
So whenever architecture code detects that cpu topology changed it might
force an update of the scheduling domains by calling arch_reinit_sched_domains
or rebuild_sched_domains.

Besides that the first patch also adds a comment to arch_update_cpu_topology
so that its purpose gets described, since most people seem to expect that
cpu core mappings are stable which isn't necessarily the case.

[PATCH/RESEND 1/2] sched: let arch_update_cpu_topology indicate if topology changed

[Heiko Carstens]

[-- Attachment #1: 001_arch_topology.diff --]
[-- Type: text/plain, Size: 2502 bytes --]

From: Heiko Carstens <heiko.carstens@de.ibm.com>

Change arch_update_cpu_topology so it returns 1 if the cpu topology changed
and 0 if it didn't change. This will be useful for the next patch which adds
a call to this function in partition_sched_domains.

Cc: Max Krasnyansky <maxk@qualcomm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
---
 arch/s390/kernel/topology.c |    5 +++--
 include/linux/topology.h    |    2 +-
 kernel/sched.c              |    8 +++++++-
 3 files changed, 11 insertions(+), 4 deletions(-)

Index: linux-2.6/kernel/sched.c
===================================================================
--- linux-2.6.orig/kernel/sched.c
+++ linux-2.6/kernel/sched.c
@@ -7712,8 +7712,14 @@ static struct sched_domain_attr *dattr_c
  */
 static cpumask_t fallback_doms;
 
-void __attribute__((weak)) arch_update_cpu_topology(void)
+/*
+ * arch_update_cpu_topology lets virtualized architectures update the
+ * cpu core maps. It is supposed to return 1 if the topology changed
+ * or 0 if it stayed the same.
+ */
+int __attribute__((weak)) arch_update_cpu_topology(void)
 {
+	return 0;
 }
 
 /*
Index: linux-2.6/arch/s390/kernel/topology.c
===================================================================
--- linux-2.6.orig/arch/s390/kernel/topology.c
+++ linux-2.6/arch/s390/kernel/topology.c
@@ -212,7 +212,7 @@ static void update_cpu_core_map(void)
 		cpu_core_map[cpu] = cpu_coregroup_map(cpu);
 }
 
-void arch_update_cpu_topology(void)
+int arch_update_cpu_topology(void)
 {
 	struct tl_info *info = tl_info;
 	struct sys_device *sysdev;
@@ -221,7 +221,7 @@ void arch_update_cpu_topology(void)
 	if (!machine_has_topology) {
 		update_cpu_core_map();
 		topology_update_polarization_simple();
-		return;
+		return 0;
 	}
 	stsi(info, 15, 1, 2);
 	tl_to_cores(info);
@@ -230,6 +230,7 @@ void arch_update_cpu_topology(void)
 		sysdev = get_cpu_sysdev(cpu);
 		kobject_uevent(&sysdev->kobj, KOBJ_CHANGE);
 	}
+	return 1;
 }
 
 static void topology_work_fn(struct work_struct *work)
Index: linux-2.6/include/linux/topology.h
===================================================================
--- linux-2.6.orig/include/linux/topology.h
+++ linux-2.6/include/linux/topology.h
@@ -49,7 +49,7 @@
 	for_each_online_node(node)			\
 		if (nr_cpus_node(node))
 
-void arch_update_cpu_topology(void);
+int arch_update_cpu_topology(void);
 
 /* Conform to ACPI 2.0 SLIT distance definitions */
 #define LOCAL_DISTANCE		10

-- 

[PATCH/RESEND 2/2] sched: re-add missing arch_update_cpu_topology call

[Heiko Carstens]

[-- Attachment #1: 002_partition_update.diff --]
[-- Type: text/plain, Size: 2260 bytes --]

From: Heiko Carstens <heiko.carstens@de.ibm.com>

arch_reinit_sched_domains used to call arch_update_cpu_topology
via arch_init_sched_domains. This call got lost with
e761b7725234276a802322549cee5255305a0930 ("cpu hotplug, sched: Introduce
cpu_active_map and redo sched domain managment (take 2)".

So we might end up with outdated and missing cpus in the cpu core
maps (architecture used to call arch_reinit_sched_domains if cpu
topology changed).

This adds a call to arch_update_cpu_topology in partition_sched_domains
which gets called whenever scheduling domains get updated. Which is
what is supposed to happen when cpu topology changes.

Since arch_update_cpu_topology returns 0 if the cpu topology didn't
change this is a NOP for all architectures but s390 (currently).

If arch_update_cpu_topology returns 1 this causes all scheduling domains
to be destroyed and rebuilt.

Cc: Max Krasnyansky <maxk@qualcomm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
---
 kernel/sched.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

Index: linux-2.6/kernel/sched.c
===================================================================
--- linux-2.6.orig/kernel/sched.c
+++ linux-2.6/kernel/sched.c
@@ -7813,17 +7813,21 @@ void partition_sched_domains(int ndoms_n
 			     struct sched_domain_attr *dattr_new)
 {
 	int i, j, n;
+	int top_changed;
 
 	mutex_lock(&sched_domains_mutex);
 
 	/* always unregister in case we don't destroy any domains */
 	unregister_sched_domain_sysctl();
 
+	/* Let architecture update cpu core mappings. */
+	top_changed = arch_update_cpu_topology();
+
 	n = doms_new ? ndoms_new : 0;
 
 	/* Destroy deleted domains */
 	for (i = 0; i < ndoms_cur; i++) {
-		for (j = 0; j < n; j++) {
+		for (j = 0; j < n && !top_changed; j++) {
 			if (cpus_equal(doms_cur[i], doms_new[j])
 			    && dattrs_equal(dattr_cur, i, dattr_new, j))
 				goto match1;
@@ -7843,7 +7847,7 @@ match1:
 
 	/* Build new domains */
 	for (i = 0; i < ndoms_new; i++) {
-		for (j = 0; j < ndoms_cur; j++) {
+		for (j = 0; j < ndoms_cur && !top_changed; j++) {
 			if (cpus_equal(doms_new[i], doms_cur[j])
 			    && dattrs_equal(dattr_new, i, dattr_cur, j))
 				goto match2;

--