[patch V2 00/67] cpuhotplug: Convert all priority notifiers to the state machine

[patch V2 00/67] cpuhotplug: Convert all priority notifiers to the state machine

[Anna-Maria Gleixner]

This series is the next step towards a state machine based hotplug machinery.

It converts all notifiers which have a priority assigned and gets rid of the
CPU_STARTING/DYING notifiers.

Changes vs. V1:
  - Addressed the review comments from Ingo, Paul and Pavel
  - Picked up acked-bys
  - Picked up the massaged changelogs from Ingo
  - Fixed the fallout from 0-day testing

Note this applies to tip/smp/hotplug. It contains the branches which are
preliminaries for that series.

Thanks,

	Anna-Maria
---
 arch/arc/kernel/time.c                        |   48 ++----
 arch/arm/kernel/smp_twd.c                     |   31 +---
 arch/arm/mach-mvebu/coherency.c               |   19 --
 arch/arm/mm/cache-l2x0.c                      |   27 +--
 arch/arm/vfp/vfpmodule.c                      |   28 ++--
 arch/arm/xen/enlighten.c                      |   41 +----
 arch/arm64/kernel/armv8_deprecated.c          |   22 ---
 arch/blackfin/kernel/perf_event.c             |   26 ---
 arch/metag/kernel/perf/perf_event.c           |   26 +--
 arch/mips/oprofile/op_model_loongson3.c       |   35 ++---
 arch/powerpc/mm/numa.c                        |   46 ++----
 arch/powerpc/perf/core-book3s.c               |   30 +---
 arch/s390/kernel/perf_cpum_cf.c               |   46 ++----
 arch/s390/kernel/perf_cpum_sf.c               |   45 ++----
 arch/sh/kernel/perf_event.c                   |   23 ---
 arch/sparc/kernel/smp_32.c                    |    2 
 arch/x86/entry/vdso/vma.c                     |   20 --
 arch/x86/events/amd/core.c                    |    6 
 arch/x86/events/amd/ibs.c                     |   64 ++++-----
 arch/x86/events/amd/power.c                   |   60 +-------
 arch/x86/events/amd/uncore.c                  |  122 ++++-------------
 arch/x86/events/core.c                        |  103 +++++++++-----
 arch/x86/events/intel/core.c                  |    4 
 arch/x86/events/intel/cqm.c                   |   49 ++-----
 arch/x86/events/intel/cstate.c                |   51 ++-----
 arch/x86/events/intel/rapl.c                  |   84 +++---------
 arch/x86/events/intel/uncore.c                |  133 ++++++-------------
 arch/x86/kernel/apb_timer.c                   |   29 +---
 arch/x86/kernel/apic/x2apic_cluster.c         |   80 ++++-------
 arch/x86/kernel/hpet.c                        |   69 ++++-----
 arch/x86/kernel/tboot.c                       |   25 +--
 arch/x86/kvm/x86.c                            |   35 +----
 arch/xtensa/kernel/perf_event.c               |   26 +--
 drivers/acpi/processor_driver.c               |   11 -
 drivers/bus/arm-cci.c                         |   53 +++----
 drivers/bus/arm-ccn.c                         |   57 ++++----
 drivers/clocksource/arm_arch_timer.c          |   54 +++----
 drivers/clocksource/arm_global_timer.c        |   39 +----
 drivers/clocksource/dummy_timer.c             |   36 -----
 drivers/clocksource/exynos_mct.c              |   46 +-----
 drivers/clocksource/metag_generic.c           |   33 ----
 drivers/clocksource/mips-gic-timer.c          |   38 +----
 drivers/clocksource/qcom-timer.c              |   41 +----
 drivers/clocksource/time-armada-370-xp.c      |   41 +----
 drivers/clocksource/timer-atlas7.c            |   41 +----
 drivers/hwtracing/coresight/coresight-etm3x.c |   90 +++++++-----
 drivers/hwtracing/coresight/coresight-etm4x.c |   87 +++++++-----
 drivers/irqchip/irq-armada-370-xp.c           |   44 ++----
 drivers/irqchip/irq-bcm2836.c                 |   34 +---
 drivers/irqchip/irq-gic-v3.c                  |   22 +--
 drivers/irqchip/irq-gic.c                     |   23 ---
 drivers/irqchip/irq-hip04.c                   |   25 ---
 drivers/leds/trigger/ledtrig-cpu.c            |   32 ++--
 drivers/perf/arm_pmu.c                        |   36 ++---
 include/linux/cpu.h                           |   23 ---
 include/linux/cpuhotplug.h                    |   78 +++++++++++
 include/linux/hrtimer.h                       |    7 +
 include/linux/perf/arm_pmu.h                  |    1 
 include/linux/perf_event.h                    |   44 +-----
 include/linux/rcutiny.h                       |    7 +
 include/linux/rcutree.h                       |    7 +
 include/linux/smp.h                           |    5 
 include/linux/timer.h                         |    6 
 include/linux/workqueue.h                     |    6 
 kernel/cpu.c                                  |   89 ++++++++----
 kernel/events/core.c                          |   56 +-------
 kernel/profile.c                              |  181 +++++++++-----------------
 kernel/rcu/tree.c                             |  105 +++++++--------
 kernel/smp.c                                  |   79 ++++-------
 kernel/time/hrtimer.c                         |   40 -----
 kernel/time/timer.c                           |   25 ---
 kernel/workqueue.c                            |  108 ++++++---------
 tools/testing/radix-tree/linux/cpu.h          |   13 -
 virt/kvm/arm/arch_timer.c                     |   35 +----
 virt/kvm/arm/vgic.c                           |   39 +----
 virt/kvm/arm/vgic/vgic-init.c                 |   31 +---
 virt/kvm/kvm_main.c                           |   32 +---
 77 files changed, 1287 insertions(+), 2068 deletions(-)

[patch V2 02/67] x86/vdso: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0001-x86-vdso-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2244 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Andy Lutomirski <luto@amacapital.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/entry/vdso/vma.c  | 20 ++++----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 5 insertions(+), 16 deletions(-)

diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c
index ab220ac..25b0368 100644
--- a/arch/x86/entry/vdso/vma.c
+++ b/arch/x86/entry/vdso/vma.c
@@ -294,15 +294,9 @@ static void vgetcpu_cpu_init(void *arg)
 	write_gdt_entry(get_cpu_gdt_table(cpu), GDT_ENTRY_PER_CPU, &d, DESCTYPE_S);
 }
 
-static int
-vgetcpu_cpu_notifier(struct notifier_block *n, unsigned long action, void *arg)
+static int vgetcpu_online(unsigned int cpu)
 {
-	long cpu = (long)arg;
-
-	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN)
-		smp_call_function_single(cpu, vgetcpu_cpu_init, NULL, 1);
-
-	return NOTIFY_DONE;
+	return smp_call_function_single(cpu, vgetcpu_cpu_init, NULL, 1);
 }
 
 static int __init init_vdso(void)
@@ -313,15 +307,9 @@ static int __init init_vdso(void)
 	init_vdso_image(&vdso_image_x32);
 #endif
 
-	cpu_notifier_register_begin();
-
-	on_each_cpu(vgetcpu_cpu_init, NULL, 1);
 	/* notifier priority > KVM */
-	__hotcpu_notifier(vgetcpu_cpu_notifier, 30);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cpuhp_setup_state(CPUHP_AP_X86_VDSO_VMA_ONLINE,
+				 "AP_X86_VDSO_VMA_ONLINE", vgetcpu_online, NULL);
 }
 subsys_initcall(init_vdso);
 #endif /* CONFIG_X86_64 */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 386374d..fe05640 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -14,6 +14,7 @@ enum cpuhp_state {
 	CPUHP_TEARDOWN_CPU,
 	CPUHP_AP_ONLINE_IDLE,
 	CPUHP_AP_SMPBOOT_THREADS,
+	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] x86/vdso: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  07d36c9e84002678d75b97d2361eb6302a718359
Gitweb:     http://git.kernel.org/tip/07d36c9e84002678d75b97d2361eb6302a718359
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:03 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:27 +0200

x86/vdso: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Andy Lutomirski <luto@amacapital.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153332.987560239@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/entry/vdso/vma.c  | 20 ++++----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 5 insertions(+), 16 deletions(-)

diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c
index ab220ac..25b0368 100644
--- a/arch/x86/entry/vdso/vma.c
+++ b/arch/x86/entry/vdso/vma.c
@@ -294,15 +294,9 @@ static void vgetcpu_cpu_init(void *arg)
 	write_gdt_entry(get_cpu_gdt_table(cpu), GDT_ENTRY_PER_CPU, &d, DESCTYPE_S);
 }
 
-static int
-vgetcpu_cpu_notifier(struct notifier_block *n, unsigned long action, void *arg)
+static int vgetcpu_online(unsigned int cpu)
 {
-	long cpu = (long)arg;
-
-	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN)
-		smp_call_function_single(cpu, vgetcpu_cpu_init, NULL, 1);
-
-	return NOTIFY_DONE;
+	return smp_call_function_single(cpu, vgetcpu_cpu_init, NULL, 1);
 }
 
 static int __init init_vdso(void)
@@ -313,15 +307,9 @@ static int __init init_vdso(void)
 	init_vdso_image(&vdso_image_x32);
 #endif
 
-	cpu_notifier_register_begin();
-
-	on_each_cpu(vgetcpu_cpu_init, NULL, 1);
 	/* notifier priority > KVM */
-	__hotcpu_notifier(vgetcpu_cpu_notifier, 30);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cpuhp_setup_state(CPUHP_AP_X86_VDSO_VMA_ONLINE,
+				 "AP_X86_VDSO_VMA_ONLINE", vgetcpu_online, NULL);
 }
 subsys_initcall(init_vdso);
 #endif /* CONFIG_X86_64 */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 386374d..fe05640 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -14,6 +14,7 @@ enum cpuhp_state {
 	CPUHP_TEARDOWN_CPU,
 	CPUHP_AP_ONLINE_IDLE,
 	CPUHP_AP_SMPBOOT_THREADS,
+	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 01/67] cpuhotplug: Handle early registration gracefully

[Anna-Maria Gleixner]

[-- Attachment #1: cpuhotplug-Handle-early-registration-gracefully.patch --]
[-- Type: text/plain, Size: 894 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

We switched the hotplug machinery to smpboot threads. Early registration of
hotplug callbacks, i.e. from do_pre_smp_initcalls(), happens before the
threads are initialized. Instead of moving the thread init, we simply handle
it in the hotplug code itself and invoke the function directly.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 kernel/cpu.c |    7 +++++++
 1 file changed, 7 insertions(+)

--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -517,6 +517,13 @@ static int cpuhp_invoke_ap_callback(int
 	if (!cpu_online(cpu))
 		return 0;
 
+	/*
+	 * If we are up and running, use the hotplug thread. For early calls
+	 * we invoke the thread function directly.
+	 */
+	if (!st->thread)
+		return cpuhp_invoke_callback(cpu, state, cb);
+
 	st->cb_state = state;
 	st->cb = cb;
 	/*

[tip:smp/hotplug] cpu/hotplug: Handle early registration gracefully

[tip-bot for Thomas Gleixner]

Commit-ID:  6a4e24518c8a10f78f44da219835239cb5aca90d
Gitweb:     http://git.kernel.org/tip/6a4e24518c8a10f78f44da219835239cb5aca90d
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:03 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:25 +0200

cpu/hotplug: Handle early registration gracefully

We switched the hotplug machinery to smpboot threads. Early registration of
hotplug callbacks, i.e. from do_pre_smp_initcalls(), happens before the
threads are initialized. Instead of moving the thread init, we simply handle
it in the hotplug code itself and invoke the function directly.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153332.896450738@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 kernel/cpu.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/kernel/cpu.c b/kernel/cpu.c
index 7b61887..fe71ce4 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -517,6 +517,13 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 	if (!cpu_online(cpu))
 		return 0;
 
+	/*
+	 * If we are up and running, use the hotplug thread. For early calls
+	 * we invoke the thread function directly.
+	 */
+	if (!st->thread)
+		return cpuhp_invoke_callback(cpu, state, cb);
+
 	st->cb_state = state;
 	st->cb = cb;
 	/*

[patch V2 03/67] irqchip/gic: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0002-irqchip-gic-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2761 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

More or less straightforward, although this driver sports some very
interesting SMP setup code. Regarding the callback ordering, this
deleted comment is interesting:

   ... the GIC needs to be up before the ARM generic timers.

That comment is half baken as the same requirement is true for perf.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/irqchip/irq-gic.c  | 23 ++++++-----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 7 insertions(+), 17 deletions(-)

diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c
index fbc4ae2..606f114 100644
--- a/drivers/irqchip/irq-gic.c
+++ b/drivers/irqchip/irq-gic.c
@@ -984,25 +984,12 @@ static int gic_irq_domain_translate(struct irq_domain *d,
 	return -EINVAL;
 }
 
-#ifdef CONFIG_SMP
-static int gic_secondary_init(struct notifier_block *nfb, unsigned long action,
-			      void *hcpu)
+static int gic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		gic_cpu_init(&gic_data[0]);
-	return NOTIFY_OK;
+	gic_cpu_init(&gic_data[0]);
+	return 0;
 }
 
-/*
- * Notifier for enabling the GIC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block gic_cpu_notifier = {
-	.notifier_call = gic_secondary_init,
-	.priority = 100,
-};
-#endif
-
 static int gic_irq_domain_alloc(struct irq_domain *domain, unsigned int virq,
 				unsigned int nr_irqs, void *arg)
 {
@@ -1148,8 +1135,10 @@ static int __init __gic_init_bases(struct gic_chip_data *gic, int irq_start,
 			gic_cpu_map[i] = 0xff;
 #ifdef CONFIG_SMP
 		set_smp_cross_call(gic_raise_softirq);
-		register_cpu_notifier(&gic_cpu_notifier);
 #endif
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_GIC_STARTING,
+					  "AP_IRQ_GIC_STARTING",
+					  gic_starting_cpu, NULL);
 		set_handle_irq(gic_handle_irq);
 		if (static_key_true(&supports_deactivate))
 			pr_info("GIC: Using split EOI/Deactivate mode\n");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index fe05640..edc8adf 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -9,6 +9,7 @@ enum cpuhp_state {
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
+	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] irqchip/gic: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  93131f7a9b5c261f9995b4a6cbd7dbbec1351d2f
Gitweb:     http://git.kernel.org/tip/93131f7a9b5c261f9995b4a6cbd7dbbec1351d2f
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:04 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:28 +0200

irqchip/gic: Convert to hotplug state machine

More or less straightforward, although this driver sports some very
interesting SMP setup code. Regarding the callback ordering, this
deleted comment is interesting:

   ... the GIC needs to be up before the ARM generic timers.

That comment is half baken as the same requirement is true for perf.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.069777215@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/irqchip/irq-gic.c  | 23 ++++++-----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 7 insertions(+), 17 deletions(-)

diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c
index fbc4ae2..606f114 100644
--- a/drivers/irqchip/irq-gic.c
+++ b/drivers/irqchip/irq-gic.c
@@ -984,25 +984,12 @@ static int gic_irq_domain_translate(struct irq_domain *d,
 	return -EINVAL;
 }
 
-#ifdef CONFIG_SMP
-static int gic_secondary_init(struct notifier_block *nfb, unsigned long action,
-			      void *hcpu)
+static int gic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		gic_cpu_init(&gic_data[0]);
-	return NOTIFY_OK;
+	gic_cpu_init(&gic_data[0]);
+	return 0;
 }
 
-/*
- * Notifier for enabling the GIC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block gic_cpu_notifier = {
-	.notifier_call = gic_secondary_init,
-	.priority = 100,
-};
-#endif
-
 static int gic_irq_domain_alloc(struct irq_domain *domain, unsigned int virq,
 				unsigned int nr_irqs, void *arg)
 {
@@ -1148,8 +1135,10 @@ static int __init __gic_init_bases(struct gic_chip_data *gic, int irq_start,
 			gic_cpu_map[i] = 0xff;
 #ifdef CONFIG_SMP
 		set_smp_cross_call(gic_raise_softirq);
-		register_cpu_notifier(&gic_cpu_notifier);
 #endif
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_GIC_STARTING,
+					  "AP_IRQ_GIC_STARTING",
+					  gic_starting_cpu, NULL);
 		set_handle_irq(gic_handle_irq);
 		if (static_key_true(&supports_deactivate))
 			pr_info("GIC: Using split EOI/Deactivate mode\n");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index fe05640..edc8adf 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -9,6 +9,7 @@ enum cpuhp_state {
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
+	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 04/67] irqchip/gicv3: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0003-irqchip-gicv3-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2381 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/irqchip/irq-gic-v3.c | 22 +++++++---------------
 include/linux/cpuhotplug.h   |  1 +
 2 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index 2c5ba0e..6fc56c3 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -538,23 +538,13 @@ static void gic_cpu_init(void)
 }
 
 #ifdef CONFIG_SMP
-static int gic_secondary_init(struct notifier_block *nfb,
-			      unsigned long action, void *hcpu)
+
+static int gic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		gic_cpu_init();
-	return NOTIFY_OK;
+	gic_cpu_init();
+	return 0;
 }
 
-/*
- * Notifier for enabling the GIC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block gic_cpu_notifier = {
-	.notifier_call = gic_secondary_init,
-	.priority = 100,
-};
-
 static u16 gic_compute_target_list(int *base_cpu, const struct cpumask *mask,
 				   unsigned long cluster_id)
 {
@@ -634,7 +624,9 @@ static void gic_raise_softirq(const struct cpumask *mask, unsigned int irq)
 static void gic_smp_init(void)
 {
 	set_smp_cross_call(gic_raise_softirq);
-	register_cpu_notifier(&gic_cpu_notifier);
+	cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_GICV3_STARTING,
+				  "AP_IRQ_GICV3_STARTING", gic_starting_cpu,
+				  NULL);
 }
 
 static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index edc8adf..d6e4a06 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -10,6 +10,7 @@ enum cpuhp_state {
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
 	CPUHP_AP_IRQ_GIC_STARTING,
+	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] irqchip/gicv3: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  6670a6d864653859229d2e921ea193a4a677dfea
Gitweb:     http://git.kernel.org/tip/6670a6d864653859229d2e921ea193a4a677dfea
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:05 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:28 +0200

irqchip/gicv3: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.163186301@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/irqchip/irq-gic-v3.c | 22 +++++++---------------
 include/linux/cpuhotplug.h   |  1 +
 2 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index 2c5ba0e..6fc56c3 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -538,23 +538,13 @@ static void gic_cpu_init(void)
 }
 
 #ifdef CONFIG_SMP
-static int gic_secondary_init(struct notifier_block *nfb,
-			      unsigned long action, void *hcpu)
+
+static int gic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		gic_cpu_init();
-	return NOTIFY_OK;
+	gic_cpu_init();
+	return 0;
 }
 
-/*
- * Notifier for enabling the GIC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block gic_cpu_notifier = {
-	.notifier_call = gic_secondary_init,
-	.priority = 100,
-};
-
 static u16 gic_compute_target_list(int *base_cpu, const struct cpumask *mask,
 				   unsigned long cluster_id)
 {
@@ -634,7 +624,9 @@ static void gic_raise_softirq(const struct cpumask *mask, unsigned int irq)
 static void gic_smp_init(void)
 {
 	set_smp_cross_call(gic_raise_softirq);
-	register_cpu_notifier(&gic_cpu_notifier);
+	cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_GICV3_STARTING,
+				  "AP_IRQ_GICV3_STARTING", gic_starting_cpu,
+				  NULL);
 }
 
 static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index edc8adf..d6e4a06 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -10,6 +10,7 @@ enum cpuhp_state {
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
 	CPUHP_AP_IRQ_GIC_STARTING,
+	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 05/67] irqchip/hip04: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0004-irqchip-hip04-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2672 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/irqchip/irq-hip04.c | 25 +++++--------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 6 insertions(+), 20 deletions(-)

diff --git a/drivers/irqchip/irq-hip04.c b/drivers/irqchip/irq-hip04.c
index 9e25d8c..021b0e0 100644
--- a/drivers/irqchip/irq-hip04.c
+++ b/drivers/irqchip/irq-hip04.c
@@ -342,26 +342,12 @@ static int hip04_irq_domain_xlate(struct irq_domain *d,
 	return ret;
 }
 
-#ifdef CONFIG_SMP
-static int hip04_irq_secondary_init(struct notifier_block *nfb,
-				    unsigned long action,
-				    void *hcpu)
+static int hip04_irq_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		hip04_irq_cpu_init(&hip04_data);
-	return NOTIFY_OK;
+	hip04_irq_cpu_init(&hip04_data);
+	return 0;
 }
 
-/*
- * Notifier for enabling the INTC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block hip04_irq_cpu_notifier = {
-	.notifier_call	= hip04_irq_secondary_init,
-	.priority	= 100,
-};
-#endif
-
 static const struct irq_domain_ops hip04_irq_domain_ops = {
 	.map	= hip04_irq_domain_map,
 	.xlate	= hip04_irq_domain_xlate,
@@ -417,13 +403,12 @@ hip04_of_init(struct device_node *node, struct device_node *parent)
 
 #ifdef CONFIG_SMP
 	set_smp_cross_call(hip04_raise_softirq);
-	register_cpu_notifier(&hip04_irq_cpu_notifier);
 #endif
 	set_handle_irq(hip04_handle_irq);
 
 	hip04_irq_dist_init(&hip04_data);
-	hip04_irq_cpu_init(&hip04_data);
-
+	cpuhp_setup_state(CPUHP_AP_IRQ_HIP04_STARTING, "AP_IRQ_HIP04_STARTING",
+			  hip04_irq_starting_cpu, NULL);
 	return 0;
 }
 IRQCHIP_DECLARE(hip04_intc, "hisilicon,hip04-intc", hip04_of_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d6e4a06..62ab4ff 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -11,6 +11,7 @@ enum cpuhp_state {
 	CPUHP_AP_SCHED_STARTING,
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
+	CPUHP_AP_IRQ_HIP04_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] irqchip/hip04: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  6c034d1736384c23ea97509e36e84f35f4a9b23c
Gitweb:     http://git.kernel.org/tip/6c034d1736384c23ea97509e36e84f35f4a9b23c
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:06 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:29 +0200

irqchip/hip04: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.244546182@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/irqchip/irq-hip04.c | 25 +++++--------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 6 insertions(+), 20 deletions(-)

diff --git a/drivers/irqchip/irq-hip04.c b/drivers/irqchip/irq-hip04.c
index 9e25d8c..021b0e0 100644
--- a/drivers/irqchip/irq-hip04.c
+++ b/drivers/irqchip/irq-hip04.c
@@ -342,26 +342,12 @@ static int hip04_irq_domain_xlate(struct irq_domain *d,
 	return ret;
 }
 
-#ifdef CONFIG_SMP
-static int hip04_irq_secondary_init(struct notifier_block *nfb,
-				    unsigned long action,
-				    void *hcpu)
+static int hip04_irq_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		hip04_irq_cpu_init(&hip04_data);
-	return NOTIFY_OK;
+	hip04_irq_cpu_init(&hip04_data);
+	return 0;
 }
 
-/*
- * Notifier for enabling the INTC CPU interface. Set an arbitrarily high
- * priority because the GIC needs to be up before the ARM generic timers.
- */
-static struct notifier_block hip04_irq_cpu_notifier = {
-	.notifier_call	= hip04_irq_secondary_init,
-	.priority	= 100,
-};
-#endif
-
 static const struct irq_domain_ops hip04_irq_domain_ops = {
 	.map	= hip04_irq_domain_map,
 	.xlate	= hip04_irq_domain_xlate,
@@ -417,13 +403,12 @@ hip04_of_init(struct device_node *node, struct device_node *parent)
 
 #ifdef CONFIG_SMP
 	set_smp_cross_call(hip04_raise_softirq);
-	register_cpu_notifier(&hip04_irq_cpu_notifier);
 #endif
 	set_handle_irq(hip04_handle_irq);
 
 	hip04_irq_dist_init(&hip04_data);
-	hip04_irq_cpu_init(&hip04_data);
-
+	cpuhp_setup_state(CPUHP_AP_IRQ_HIP04_STARTING, "AP_IRQ_HIP04_STARTING",
+			  hip04_irq_starting_cpu, NULL);
 	return 0;
 }
 IRQCHIP_DECLARE(hip04_intc, "hisilicon,hip04-intc", hip04_of_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d6e4a06..62ab4ff 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -11,6 +11,7 @@ enum cpuhp_state {
 	CPUHP_AP_SCHED_STARTING,
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
+	CPUHP_AP_IRQ_HIP04_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 07/67] irqchip/bcm2836: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0006-irqchip-bcm2836-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2884 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/irqchip/irq-bcm2836.c | 34 +++++++++++++---------------------
 include/linux/cpuhotplug.h    |  1 +
 2 files changed, 14 insertions(+), 21 deletions(-)

diff --git a/drivers/irqchip/irq-bcm2836.c b/drivers/irqchip/irq-bcm2836.c
index 72ff1d5..f2575cb 100644
--- a/drivers/irqchip/irq-bcm2836.c
+++ b/drivers/irqchip/irq-bcm2836.c
@@ -202,26 +202,19 @@ static void bcm2836_arm_irqchip_send_ipi(const struct cpumask *mask,
 	}
 }
 
-/* Unmasks the IPI on the CPU when it's online. */
-static int bcm2836_arm_irqchip_cpu_notify(struct notifier_block *nfb,
-					  unsigned long action, void *hcpu)
+static int bcm2836_cpu_starting(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-	unsigned int int_reg = LOCAL_MAILBOX_INT_CONTROL0;
-	unsigned int mailbox = 0;
-
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		bcm2836_arm_irqchip_unmask_per_cpu_irq(int_reg, mailbox, cpu);
-	else if (action == CPU_DYING)
-		bcm2836_arm_irqchip_mask_per_cpu_irq(int_reg, mailbox, cpu);
-
-	return NOTIFY_OK;
+	bcm2836_arm_irqchip_unmask_per_cpu_irq(LOCAL_MAILBOX_INT_CONTROL0, 0,
+					       cpu);
+	return 0;
 }
 
-static struct notifier_block bcm2836_arm_irqchip_cpu_notifier = {
-	.notifier_call = bcm2836_arm_irqchip_cpu_notify,
-	.priority = 100,
-};
+static int bcm2836_cpu_dying(unsigned int cpu)
+{
+	bcm2836_arm_irqchip_mask_per_cpu_irq(LOCAL_MAILBOX_INT_CONTROL0, 0,
+					     cpu);
+	return 0;
+}
 
 #ifdef CONFIG_ARM
 int __init bcm2836_smp_boot_secondary(unsigned int cpu,
@@ -251,10 +244,9 @@ bcm2836_arm_irqchip_smp_init(void)
 {
 #ifdef CONFIG_SMP
 	/* Unmask IPIs to the boot CPU. */
-	bcm2836_arm_irqchip_cpu_notify(&bcm2836_arm_irqchip_cpu_notifier,
-				       CPU_STARTING,
-				       (void *)(uintptr_t)smp_processor_id());
-	register_cpu_notifier(&bcm2836_arm_irqchip_cpu_notifier);
+	cpuhp_setup_state(CPUHP_AP_IRQ_BCM2836_STARTING,
+			  "AP_IRQ_BCM2836_STARTING", bcm2836_cpu_starting,
+			  bcm2836_cpu_dying);
 
 	set_smp_cross_call(bcm2836_arm_irqchip_send_ipi);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 3a08ab9..420cfcf 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -14,6 +14,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_HIP04_STARTING,
 	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
+	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] irqchip/bcm2836: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  7ca04bc277431a271061dd85f98b6da4171ad2e4
Gitweb:     http://git.kernel.org/tip/7ca04bc277431a271061dd85f98b6da4171ad2e4
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:07 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:30 +0200

irqchip/bcm2836: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.416260485@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/irqchip/irq-bcm2836.c | 34 +++++++++++++---------------------
 include/linux/cpuhotplug.h    |  1 +
 2 files changed, 14 insertions(+), 21 deletions(-)

diff --git a/drivers/irqchip/irq-bcm2836.c b/drivers/irqchip/irq-bcm2836.c
index 72ff1d5..f2575cb 100644
--- a/drivers/irqchip/irq-bcm2836.c
+++ b/drivers/irqchip/irq-bcm2836.c
@@ -202,26 +202,19 @@ static void bcm2836_arm_irqchip_send_ipi(const struct cpumask *mask,
 	}
 }
 
-/* Unmasks the IPI on the CPU when it's online. */
-static int bcm2836_arm_irqchip_cpu_notify(struct notifier_block *nfb,
-					  unsigned long action, void *hcpu)
+static int bcm2836_cpu_starting(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-	unsigned int int_reg = LOCAL_MAILBOX_INT_CONTROL0;
-	unsigned int mailbox = 0;
-
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		bcm2836_arm_irqchip_unmask_per_cpu_irq(int_reg, mailbox, cpu);
-	else if (action == CPU_DYING)
-		bcm2836_arm_irqchip_mask_per_cpu_irq(int_reg, mailbox, cpu);
-
-	return NOTIFY_OK;
+	bcm2836_arm_irqchip_unmask_per_cpu_irq(LOCAL_MAILBOX_INT_CONTROL0, 0,
+					       cpu);
+	return 0;
 }
 
-static struct notifier_block bcm2836_arm_irqchip_cpu_notifier = {
-	.notifier_call = bcm2836_arm_irqchip_cpu_notify,
-	.priority = 100,
-};
+static int bcm2836_cpu_dying(unsigned int cpu)
+{
+	bcm2836_arm_irqchip_mask_per_cpu_irq(LOCAL_MAILBOX_INT_CONTROL0, 0,
+					     cpu);
+	return 0;
+}
 
 #ifdef CONFIG_ARM
 int __init bcm2836_smp_boot_secondary(unsigned int cpu,
@@ -251,10 +244,9 @@ bcm2836_arm_irqchip_smp_init(void)
 {
 #ifdef CONFIG_SMP
 	/* Unmask IPIs to the boot CPU. */
-	bcm2836_arm_irqchip_cpu_notify(&bcm2836_arm_irqchip_cpu_notifier,
-				       CPU_STARTING,
-				       (void *)(uintptr_t)smp_processor_id());
-	register_cpu_notifier(&bcm2836_arm_irqchip_cpu_notifier);
+	cpuhp_setup_state(CPUHP_AP_IRQ_BCM2836_STARTING,
+			  "AP_IRQ_BCM2836_STARTING", bcm2836_cpu_starting,
+			  bcm2836_cpu_dying);
 
 	set_smp_cross_call(bcm2836_arm_irqchip_send_ipi);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 3a08ab9..420cfcf 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -14,6 +14,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_HIP04_STARTING,
 	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
+	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 06/67] irqchip/armada-370-xp: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0005-irqchip-armada-370-xp-Convert-to-hotplug-state-machi.patch --]
[-- Type: text/plain, Size: 3460 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/irqchip/irq-armada-370-xp.c | 44 +++++++++++++------------------------
 include/linux/cpuhotplug.h          |  2 ++
 2 files changed, 17 insertions(+), 29 deletions(-)

diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c
index e7dc6cb..e2b8fee 100644
--- a/drivers/irqchip/irq-armada-370-xp.c
+++ b/drivers/irqchip/irq-armada-370-xp.c
@@ -344,40 +344,22 @@ static void armada_mpic_send_doorbell(const struct cpumask *mask,
 	writel((map << 8) | irq, main_int_base +
 		ARMADA_370_XP_SW_TRIG_INT_OFFS);
 }
+#endif
 
-static int armada_xp_mpic_secondary_init(struct notifier_block *nfb,
-					 unsigned long action, void *hcpu)
+static int armada_xp_mpic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN) {
-		armada_xp_mpic_perf_init();
-		armada_xp_mpic_smp_cpu_init();
-	}
-
-	return NOTIFY_OK;
+	armada_xp_mpic_perf_init();
+	armada_xp_mpic_smp_cpu_init();
+	return 0;
 }
 
-static struct notifier_block armada_370_xp_mpic_cpu_notifier = {
-	.notifier_call = armada_xp_mpic_secondary_init,
-	.priority = 100,
-};
-
-static int mpic_cascaded_secondary_init(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int mpic_cascaded_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN) {
-		armada_xp_mpic_perf_init();
-		enable_percpu_irq(parent_irq, IRQ_TYPE_NONE);
-	}
-
-	return NOTIFY_OK;
+	armada_xp_mpic_perf_init();
+	enable_percpu_irq(parent_irq, IRQ_TYPE_NONE);
+	return 0;
 }
 
-static struct notifier_block mpic_cascaded_cpu_notifier = {
-	.notifier_call = mpic_cascaded_secondary_init,
-	.priority = 100,
-};
-#endif /* CONFIG_SMP */
-
 static const struct irq_domain_ops armada_370_xp_mpic_irq_ops = {
 	.map = armada_370_xp_mpic_irq_map,
 	.xlate = irq_domain_xlate_onecell,
@@ -595,11 +577,15 @@ static int __init armada_370_xp_mpic_of_init(struct device_node *node,
 		set_handle_irq(armada_370_xp_handle_irq);
 #ifdef CONFIG_SMP
 		set_smp_cross_call(armada_mpic_send_doorbell);
-		register_cpu_notifier(&armada_370_xp_mpic_cpu_notifier);
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_ARMADA_XP_STARTING,
+					  "AP_IRQ_ARMADA_XP_STARTING",
+					  armada_xp_mpic_starting_cpu, NULL);
 #endif
 	} else {
 #ifdef CONFIG_SMP
-		register_cpu_notifier(&mpic_cascaded_cpu_notifier);
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
+					  "AP_IRQ_ARMADA_CASC_STARTING",
+					  mpic_cascaded_starting_cpu, NULL);
 #endif
 		irq_set_chained_handler(parent_irq,
 					armada_370_xp_mpic_handle_cascade_irq);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 62ab4ff..3a08ab9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,8 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_IRQ_HIP04_STARTING,
+	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
+	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] irqchip/armada-370-xp: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  cb5ff2d245c1023f1004eca73038b2cae87bcc6a
Gitweb:     http://git.kernel.org/tip/cb5ff2d245c1023f1004eca73038b2cae87bcc6a
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:07 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:29 +0200

irqchip/armada-370-xp: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.330661455@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/irqchip/irq-armada-370-xp.c | 44 +++++++++++++------------------------
 include/linux/cpuhotplug.h          |  2 ++
 2 files changed, 17 insertions(+), 29 deletions(-)

diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c
index e7dc6cb..e2b8fee 100644
--- a/drivers/irqchip/irq-armada-370-xp.c
+++ b/drivers/irqchip/irq-armada-370-xp.c
@@ -344,40 +344,22 @@ static void armada_mpic_send_doorbell(const struct cpumask *mask,
 	writel((map << 8) | irq, main_int_base +
 		ARMADA_370_XP_SW_TRIG_INT_OFFS);
 }
+#endif
 
-static int armada_xp_mpic_secondary_init(struct notifier_block *nfb,
-					 unsigned long action, void *hcpu)
+static int armada_xp_mpic_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN) {
-		armada_xp_mpic_perf_init();
-		armada_xp_mpic_smp_cpu_init();
-	}
-
-	return NOTIFY_OK;
+	armada_xp_mpic_perf_init();
+	armada_xp_mpic_smp_cpu_init();
+	return 0;
 }
 
-static struct notifier_block armada_370_xp_mpic_cpu_notifier = {
-	.notifier_call = armada_xp_mpic_secondary_init,
-	.priority = 100,
-};
-
-static int mpic_cascaded_secondary_init(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int mpic_cascaded_starting_cpu(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN) {
-		armada_xp_mpic_perf_init();
-		enable_percpu_irq(parent_irq, IRQ_TYPE_NONE);
-	}
-
-	return NOTIFY_OK;
+	armada_xp_mpic_perf_init();
+	enable_percpu_irq(parent_irq, IRQ_TYPE_NONE);
+	return 0;
 }
 
-static struct notifier_block mpic_cascaded_cpu_notifier = {
-	.notifier_call = mpic_cascaded_secondary_init,
-	.priority = 100,
-};
-#endif /* CONFIG_SMP */
-
 static const struct irq_domain_ops armada_370_xp_mpic_irq_ops = {
 	.map = armada_370_xp_mpic_irq_map,
 	.xlate = irq_domain_xlate_onecell,
@@ -595,11 +577,15 @@ static int __init armada_370_xp_mpic_of_init(struct device_node *node,
 		set_handle_irq(armada_370_xp_handle_irq);
 #ifdef CONFIG_SMP
 		set_smp_cross_call(armada_mpic_send_doorbell);
-		register_cpu_notifier(&armada_370_xp_mpic_cpu_notifier);
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_ARMADA_XP_STARTING,
+					  "AP_IRQ_ARMADA_XP_STARTING",
+					  armada_xp_mpic_starting_cpu, NULL);
 #endif
 	} else {
 #ifdef CONFIG_SMP
-		register_cpu_notifier(&mpic_cascaded_cpu_notifier);
+		cpuhp_setup_state_nocalls(CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
+					  "AP_IRQ_ARMADA_CASC_STARTING",
+					  mpic_cascaded_starting_cpu, NULL);
 #endif
 		irq_set_chained_handler(parent_irq,
 					armada_370_xp_mpic_handle_cascade_irq);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 62ab4ff..3a08ab9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,8 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_IRQ_HIP04_STARTING,
+	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
+	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 08/67] ARM: mvebu: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0007-ARM-mvebu-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2476 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Gregory Clement <gregory.clement@free-electrons.com>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm/mach-mvebu/coherency.c | 19 ++++++-------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 7 insertions(+), 13 deletions(-)

diff --git a/arch/arm/mach-mvebu/coherency.c b/arch/arm/mach-mvebu/coherency.c
index 7e989d6..77aaa52 100644
--- a/arch/arm/mach-mvebu/coherency.c
+++ b/arch/arm/mach-mvebu/coherency.c
@@ -111,20 +111,12 @@ static struct notifier_block mvebu_hwcc_pci_nb __maybe_unused = {
 	.notifier_call = mvebu_hwcc_notifier,
 };
 
-static int armada_xp_clear_shared_l2_notifier_func(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int armada_xp_clear_l2_starting(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		armada_xp_clear_shared_l2();
-
-	return NOTIFY_OK;
+	armada_xp_clear_shared_l2();
+	return 0;
 }
 
-static struct notifier_block armada_xp_clear_shared_l2_notifier = {
-	.notifier_call = armada_xp_clear_shared_l2_notifier_func,
-	.priority = 100,
-};
-
 static void __init armada_370_coherency_init(struct device_node *np)
 {
 	struct resource res;
@@ -155,8 +147,9 @@ static void __init armada_370_coherency_init(struct device_node *np)
 
 	of_node_put(cpu_config_np);
 
-	register_cpu_notifier(&armada_xp_clear_shared_l2_notifier);
-
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_MVEBU_COHERENCY,
+				  "AP_ARM_MVEBU_COHERENCY",
+				  armada_xp_clear_l2_starting, NULL);
 exit:
 	set_cpu_coherent();
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 420cfcf..d769ec9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -15,6 +15,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
+	CPUHP_AP_ARM_MVEBU_COHERENCY,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] ARM/mvebu: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  7fbbaebf8cd18cca58e84837fd80c07ea8d8bd56
Gitweb:     http://git.kernel.org/tip/7fbbaebf8cd18cca58e84837fd80c07ea8d8bd56
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:08 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:30 +0200

ARM/mvebu: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Gregory Clement <gregory.clement@free-electrons.com>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.503198935@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm/mach-mvebu/coherency.c | 19 ++++++-------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 7 insertions(+), 13 deletions(-)

diff --git a/arch/arm/mach-mvebu/coherency.c b/arch/arm/mach-mvebu/coherency.c
index 7e989d6..77aaa52 100644
--- a/arch/arm/mach-mvebu/coherency.c
+++ b/arch/arm/mach-mvebu/coherency.c
@@ -111,20 +111,12 @@ static struct notifier_block mvebu_hwcc_pci_nb __maybe_unused = {
 	.notifier_call = mvebu_hwcc_notifier,
 };
 
-static int armada_xp_clear_shared_l2_notifier_func(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int armada_xp_clear_l2_starting(unsigned int cpu)
 {
-	if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		armada_xp_clear_shared_l2();
-
-	return NOTIFY_OK;
+	armada_xp_clear_shared_l2();
+	return 0;
 }
 
-static struct notifier_block armada_xp_clear_shared_l2_notifier = {
-	.notifier_call = armada_xp_clear_shared_l2_notifier_func,
-	.priority = 100,
-};
-
 static void __init armada_370_coherency_init(struct device_node *np)
 {
 	struct resource res;
@@ -155,8 +147,9 @@ static void __init armada_370_coherency_init(struct device_node *np)
 
 	of_node_put(cpu_config_np);
 
-	register_cpu_notifier(&armada_xp_clear_shared_l2_notifier);
-
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_MVEBU_COHERENCY,
+				  "AP_ARM_MVEBU_COHERENCY",
+				  armada_xp_clear_l2_starting, NULL);
 exit:
 	set_cpu_coherent();
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 420cfcf..d769ec9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -15,6 +15,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_XP_STARTING,
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
+	CPUHP_AP_ARM_MVEBU_COHERENCY,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 09/67] perf/core: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0008-perf-core-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 5785 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Actually a nice symmetric startup/teardown pair which fits properly into
the state machine concept. In the long run we should be able to invoke
the startup callback for the boot CPU via the state machine and get
rid of the init function which invokes it on the boot CPU.

Note: This comes actually before the perf hardware callbacks. In the notifier
model the hardware callbacks have a higher priority than the core
callback. But that's solely for CPU offline so that hardware migration of
events happens before the core is notified about the outgoing CPU.

With the symetric state array model we have the following ordering:

 UP:     core -> hardware
 DOWN:   hardware -> core

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |    2 +
 include/linux/perf_event.h |    9 +++++++
 kernel/cpu.c               |   11 ++++++++
 kernel/events/core.c       |   56 ++++++++-------------------------------------
 4 files changed, 32 insertions(+), 46 deletions(-)

--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -4,6 +4,7 @@
 enum cpuhp_state {
 	CPUHP_OFFLINE,
 	CPUHP_CREATE_THREADS,
+	CPUHP_PERF_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -22,6 +23,7 @@ enum cpuhp_state {
 	CPUHP_AP_ONLINE_IDLE,
 	CPUHP_AP_SMPBOOT_THREADS,
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
+	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1354,4 +1354,13 @@ static ssize_t								\
 									\
 static struct device_attribute format_attr_##_name = __ATTR_RO(_name)
 
+/* Performance counter hotplug functions */
+#ifdef CONFIG_PERF_EVENTS
+int perf_event_init_cpu(unsigned int cpu);
+int perf_event_exit_cpu(unsigned int cpu);
+#else
+#define perf_event_init_cpu	NULL
+#define perf_event_exit_cpu	NULL
+#endif
+
 #endif /* _LINUX_PERF_EVENT_H */
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1180,6 +1180,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.teardown		= NULL,
 		.cant_stop		= true,
 	},
+	[CPUHP_PERF_PREPARE] = {
+		.name = "perf prepare",
+		.startup = perf_event_init_cpu,
+		.teardown = perf_event_exit_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1255,6 +1260,12 @@ static struct cpuhp_step cpuhp_ap_states
 		.startup		= smpboot_unpark_threads,
 		.teardown		= NULL,
 	},
+	[CPUHP_AP_PERF_ONLINE] = {
+		.name = "perf online",
+		.startup = perf_event_init_cpu,
+		.teardown = perf_event_exit_cpu,
+	},
+
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
 	 * are converted to states.
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -10255,7 +10255,7 @@ static void __init perf_event_init_all_c
 	}
 }
 
-static void perf_event_init_cpu(int cpu)
+int perf_event_init_cpu(unsigned int cpu)
 {
 	struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);
 
@@ -10268,6 +10268,7 @@ static void perf_event_init_cpu(int cpu)
 		rcu_assign_pointer(swhash->swevent_hlist, hlist);
 	}
 	mutex_unlock(&swhash->hlist_mutex);
+	return 0;
 }
 
 #if defined CONFIG_HOTPLUG_CPU || defined CONFIG_KEXEC_CORE
@@ -10299,14 +10300,17 @@ static void perf_event_exit_cpu_context(
 	}
 	srcu_read_unlock(&pmus_srcu, idx);
 }
+#else
+
+static void perf_event_exit_cpu_context(int cpu) { }
+
+#endif
 
-static void perf_event_exit_cpu(int cpu)
+int perf_event_exit_cpu(unsigned int cpu)
 {
 	perf_event_exit_cpu_context(cpu);
+	return 0;
 }
-#else
-static inline void perf_event_exit_cpu(int cpu) { }
-#endif
 
 static int
 perf_reboot(struct notifier_block *notifier, unsigned long val, void *v)
@@ -10328,46 +10332,6 @@ static struct notifier_block perf_reboot
 	.priority = INT_MIN,
 };
 
-static int
-perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-
-	case CPU_UP_PREPARE:
-		/*
-		 * This must be done before the CPU comes alive, because the
-		 * moment we can run tasks we can encounter (software) events.
-		 *
-		 * Specifically, someone can have inherited events on kthreadd
-		 * or a pre-existing worker thread that gets re-bound.
-		 */
-		perf_event_init_cpu(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		/*
-		 * This must be done before the CPU dies because after that an
-		 * active event might want to IPI the CPU and that'll not work
-		 * so great for dead CPUs.
-		 *
-		 * XXX smp_call_function_single() return -ENXIO without a warn
-		 * so we could possibly deal with this.
-		 *
-		 * This is safe against new events arriving because
-		 * sys_perf_event_open() serializes against hotplug using
-		 * get_online_cpus().
-		 */
-		perf_event_exit_cpu(cpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
 void __init perf_event_init(void)
 {
 	int ret;
@@ -10380,7 +10344,7 @@ void __init perf_event_init(void)
 	perf_pmu_register(&perf_cpu_clock, NULL, -1);
 	perf_pmu_register(&perf_task_clock, NULL, -1);
 	perf_tp_register();
-	perf_cpu_notifier(perf_cpu_notify);
+	perf_event_init_cpu(smp_processor_id());
 	register_reboot_notifier(&perf_reboot_notifier);
 
 	ret = init_hw_breakpoint();

[tip:smp/hotplug] perf/core: Convert to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  00e16c3d68fce504e880f59c9bdf23b2a4759d6d
Gitweb:     http://git.kernel.org/tip/00e16c3d68fce504e880f59c9bdf23b2a4759d6d
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:09 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:31 +0200

perf/core: Convert to hotplug state machine

Actually a nice symmetric startup/teardown pair which fits properly into
the state machine concept. In the long run we should be able to invoke
the startup callback for the boot CPU via the state machine and get
rid of the init function which invokes it on the boot CPU.

Note: This comes actually before the perf hardware callbacks. In the notifier
model the hardware callbacks have a higher priority than the core
callback. But that's solely for CPU offline so that hardware migration of
events happens before the core is notified about the outgoing CPU.

With the symetric state array model we have the following ordering:

 UP:     core -> hardware
 DOWN:   hardware -> core

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.587514098@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  2 ++
 include/linux/perf_event.h |  9 ++++++++
 kernel/cpu.c               | 11 +++++++++
 kernel/events/core.c       | 56 +++++++++-------------------------------------
 4 files changed, 32 insertions(+), 46 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d769ec9..067082e 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -4,6 +4,7 @@
 enum cpuhp_state {
 	CPUHP_OFFLINE,
 	CPUHP_CREATE_THREADS,
+	CPUHP_PERF_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -22,6 +23,7 @@ enum cpuhp_state {
 	CPUHP_AP_ONLINE_IDLE,
 	CPUHP_AP_SMPBOOT_THREADS,
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
+	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 1a827ce..9abeb69 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1354,4 +1354,13 @@ _name##_show(struct device *dev,					\
 									\
 static struct device_attribute format_attr_##_name = __ATTR_RO(_name)
 
+/* Performance counter hotplug functions */
+#ifdef CONFIG_PERF_EVENTS
+int perf_event_init_cpu(unsigned int cpu);
+int perf_event_exit_cpu(unsigned int cpu);
+#else
+#define perf_event_init_cpu	NULL
+#define perf_event_exit_cpu	NULL
+#endif
+
 #endif /* _LINUX_PERF_EVENT_H */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index fe71ce4..3705d90 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1180,6 +1180,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.teardown		= NULL,
 		.cant_stop		= true,
 	},
+	[CPUHP_PERF_PREPARE] = {
+		.name = "perf prepare",
+		.startup = perf_event_init_cpu,
+		.teardown = perf_event_exit_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1257,6 +1262,12 @@ static struct cpuhp_step cpuhp_ap_states[] = {
 		.startup		= smpboot_unpark_threads,
 		.teardown		= NULL,
 	},
+	[CPUHP_AP_PERF_ONLINE] = {
+		.name = "perf online",
+		.startup = perf_event_init_cpu,
+		.teardown = perf_event_exit_cpu,
+	},
+
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
 	 * are converted to states.
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 43d43a2d..f3ef1c2 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -10255,7 +10255,7 @@ static void __init perf_event_init_all_cpus(void)
 	}
 }
 
-static void perf_event_init_cpu(int cpu)
+int perf_event_init_cpu(unsigned int cpu)
 {
 	struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);
 
@@ -10268,6 +10268,7 @@ static void perf_event_init_cpu(int cpu)
 		rcu_assign_pointer(swhash->swevent_hlist, hlist);
 	}
 	mutex_unlock(&swhash->hlist_mutex);
+	return 0;
 }
 
 #if defined CONFIG_HOTPLUG_CPU || defined CONFIG_KEXEC_CORE
@@ -10299,14 +10300,17 @@ static void perf_event_exit_cpu_context(int cpu)
 	}
 	srcu_read_unlock(&pmus_srcu, idx);
 }
+#else
+
+static void perf_event_exit_cpu_context(int cpu) { }
+
+#endif
 
-static void perf_event_exit_cpu(int cpu)
+int perf_event_exit_cpu(unsigned int cpu)
 {
 	perf_event_exit_cpu_context(cpu);
+	return 0;
 }
-#else
-static inline void perf_event_exit_cpu(int cpu) { }
-#endif
 
 static int
 perf_reboot(struct notifier_block *notifier, unsigned long val, void *v)
@@ -10328,46 +10332,6 @@ static struct notifier_block perf_reboot_notifier = {
 	.priority = INT_MIN,
 };
 
-static int
-perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-
-	case CPU_UP_PREPARE:
-		/*
-		 * This must be done before the CPU comes alive, because the
-		 * moment we can run tasks we can encounter (software) events.
-		 *
-		 * Specifically, someone can have inherited events on kthreadd
-		 * or a pre-existing worker thread that gets re-bound.
-		 */
-		perf_event_init_cpu(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		/*
-		 * This must be done before the CPU dies because after that an
-		 * active event might want to IPI the CPU and that'll not work
-		 * so great for dead CPUs.
-		 *
-		 * XXX smp_call_function_single() return -ENXIO without a warn
-		 * so we could possibly deal with this.
-		 *
-		 * This is safe against new events arriving because
-		 * sys_perf_event_open() serializes against hotplug using
-		 * get_online_cpus().
-		 */
-		perf_event_exit_cpu(cpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
 void __init perf_event_init(void)
 {
 	int ret;
@@ -10380,7 +10344,7 @@ void __init perf_event_init(void)
 	perf_pmu_register(&perf_cpu_clock, NULL, -1);
 	perf_pmu_register(&perf_task_clock, NULL, -1);
 	perf_tp_register();
-	perf_cpu_notifier(perf_cpu_notify);
+	perf_event_init_cpu(smp_processor_id());
 	register_reboot_notifier(&perf_reboot_notifier);
 
 	ret = init_hw_breakpoint();

[patch V2 10/67] perf/x86: Convert the core to the hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0009-perf-x86-Convert-the-core-to-the-hotplug-state-machi.patch --]
[-- Type: text/plain, Size: 6344 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Replace the perf_notifier() install mechanism, which invokes magically
the callback on the current CPU. Convert the hardware specific
callbacks which are invoked from the x86 perf core to return proper
error codes instead of totally pointless NOTIFY_BAD return values.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Adam Borowski <kilobyte@angband.pl>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/amd/core.c   |   6 +--
 arch/x86/events/core.c       | 103 +++++++++++++++++++++++++++----------------
 arch/x86/events/intel/core.c |   4 +-
 include/linux/cpuhotplug.h   |   3 ++
 4 files changed, 74 insertions(+), 42 deletions(-)

diff --git a/arch/x86/events/amd/core.c b/arch/x86/events/amd/core.c
index bd3e842..e07a22b 100644
--- a/arch/x86/events/amd/core.c
+++ b/arch/x86/events/amd/core.c
@@ -370,13 +370,13 @@ static int amd_pmu_cpu_prepare(int cpu)
 	WARN_ON_ONCE(cpuc->amd_nb);
 
 	if (!x86_pmu.amd_nb_constraints)
-		return NOTIFY_OK;
+		return 0;
 
 	cpuc->amd_nb = amd_alloc_nb(cpu);
 	if (!cpuc->amd_nb)
-		return NOTIFY_BAD;
+		return -ENOMEM;
 
-	return NOTIFY_OK;
+	return 0;
 }
 
 static void amd_pmu_cpu_starting(int cpu)
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 26ced53..4ce3745 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -1477,49 +1477,49 @@ NOKPROBE_SYMBOL(perf_event_nmi_handler);
 struct event_constraint emptyconstraint;
 struct event_constraint unconstrained;
 
-static int
-x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
+static int x86_pmu_prepare_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (long)hcpu;
 	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
-	int i, ret = NOTIFY_OK;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		for (i = 0 ; i < X86_PERF_KFREE_MAX; i++)
-			cpuc->kfree_on_online[i] = NULL;
-		if (x86_pmu.cpu_prepare)
-			ret = x86_pmu.cpu_prepare(cpu);
-		break;
-
-	case CPU_STARTING:
-		if (x86_pmu.cpu_starting)
-			x86_pmu.cpu_starting(cpu);
-		break;
+	int i;
 
-	case CPU_ONLINE:
-		for (i = 0 ; i < X86_PERF_KFREE_MAX; i++) {
-			kfree(cpuc->kfree_on_online[i]);
-			cpuc->kfree_on_online[i] = NULL;
-		}
-		break;
+	for (i = 0 ; i < X86_PERF_KFREE_MAX; i++)
+		cpuc->kfree_on_online[i] = NULL;
+	if (x86_pmu.cpu_prepare)
+		return x86_pmu.cpu_prepare(cpu);
+	return 0;
+}
 
-	case CPU_DYING:
-		if (x86_pmu.cpu_dying)
-			x86_pmu.cpu_dying(cpu);
-		break;
+static int x86_pmu_dead_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_dead)
+		x86_pmu.cpu_dead(cpu);
+	return 0;
+}
 
-	case CPU_UP_CANCELED:
-	case CPU_DEAD:
-		if (x86_pmu.cpu_dead)
-			x86_pmu.cpu_dead(cpu);
-		break;
+static int x86_pmu_online_cpu(unsigned int cpu)
+{
+	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
+	int i;
 
-	default:
-		break;
+	for (i = 0 ; i < X86_PERF_KFREE_MAX; i++) {
+		kfree(cpuc->kfree_on_online[i]);
+		cpuc->kfree_on_online[i] = NULL;
 	}
+	return 0;
+}
 
-	return ret;
+static int x86_pmu_starting_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_starting)
+		x86_pmu.cpu_starting(cpu);
+	return 0;
+}
+
+static int x86_pmu_dying_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_dying)
+		x86_pmu.cpu_dying(cpu);
+	return 0;
 }
 
 static void __init pmu_check_apic(void)
@@ -1764,10 +1764,39 @@ static int __init init_hw_perf_events(void)
 	pr_info("... fixed-purpose events:   %d\n",     x86_pmu.num_counters_fixed);
 	pr_info("... event mask:             %016Lx\n", x86_pmu.intel_ctrl);
 
-	perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(x86_pmu_notifier);
+	/*
+	 * Install callbacks. Core will call them for each online
+	 * cpu.
+	 */
+	err = cpuhp_setup_state(CPUHP_PERF_X86_PREPARE, "PERF_X86_PREPARE",
+				x86_pmu_prepare_cpu, x86_pmu_dead_cpu);
+	if (err)
+		return err;
+
+	err = cpuhp_setup_state(CPUHP_AP_PERF_X86_STARTING,
+				"AP_PERF_X86_STARTING", x86_pmu_starting_cpu,
+				x86_pmu_dying_cpu);
+	if (err)
+		goto out;
+
+	err = cpuhp_setup_state(CPUHP_AP_PERF_X86_ONLINE, "AP_PERF_X86_ONLINE",
+				x86_pmu_online_cpu, NULL);
+	if (err)
+		goto out1;
+
+	err = perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
+	if (err)
+		goto out2;
 
 	return 0;
+
+out2:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_ONLINE);
+out1:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_STARTING);
+out:
+	cpuhp_remove_state(CPUHP_PERF_X86_PREPARE);
+	return err;
 }
 early_initcall(init_hw_perf_events);
 
diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index 9b4f9d3..6a1441b 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -3050,7 +3050,7 @@ static int intel_pmu_cpu_prepare(int cpu)
 		cpuc->excl_thread_id = 0;
 	}
 
-	return NOTIFY_OK;
+	return 0;
 
 err_constraint_list:
 	kfree(cpuc->constraint_list);
@@ -3061,7 +3061,7 @@ err_shared_regs:
 	cpuc->shared_regs = NULL;
 
 err:
-	return NOTIFY_BAD;
+	return -ENOMEM;
 }
 
 static void intel_pmu_cpu_starting(int cpu)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 067082e..858915f 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -5,6 +5,7 @@ enum cpuhp_state {
 	CPUHP_OFFLINE,
 	CPUHP_CREATE_THREADS,
 	CPUHP_PERF_PREPARE,
+	CPUHP_PERF_X86_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -17,6 +18,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
+	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -24,6 +26,7 @@ enum cpuhp_state {
 	CPUHP_AP_SMPBOOT_THREADS,
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_PERF_ONLINE,
+	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86: Convert the core to the hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  95ca792c7582fde2c435f84cbe872616e3c951f6
Gitweb:     http://git.kernel.org/tip/95ca792c7582fde2c435f84cbe872616e3c951f6
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:10 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:32 +0200

perf/x86: Convert the core to the hotplug state machine

Replace the perf_notifier() install mechanism, which invokes magically
the callback on the current CPU. Convert the hardware specific
callbacks which are invoked from the x86 perf core to return proper
error codes instead of totally pointless NOTIFY_BAD return values.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Adam Borowski <kilobyte@angband.pl>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.670720553@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/amd/core.c   |   6 +--
 arch/x86/events/core.c       | 103 +++++++++++++++++++++++++++----------------
 arch/x86/events/intel/core.c |   4 +-
 include/linux/cpuhotplug.h   |   3 ++
 4 files changed, 74 insertions(+), 42 deletions(-)

diff --git a/arch/x86/events/amd/core.c b/arch/x86/events/amd/core.c
index bd3e842..e07a22b 100644
--- a/arch/x86/events/amd/core.c
+++ b/arch/x86/events/amd/core.c
@@ -370,13 +370,13 @@ static int amd_pmu_cpu_prepare(int cpu)
 	WARN_ON_ONCE(cpuc->amd_nb);
 
 	if (!x86_pmu.amd_nb_constraints)
-		return NOTIFY_OK;
+		return 0;
 
 	cpuc->amd_nb = amd_alloc_nb(cpu);
 	if (!cpuc->amd_nb)
-		return NOTIFY_BAD;
+		return -ENOMEM;
 
-	return NOTIFY_OK;
+	return 0;
 }
 
 static void amd_pmu_cpu_starting(int cpu)
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 26ced53..4ce3745 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -1477,49 +1477,49 @@ NOKPROBE_SYMBOL(perf_event_nmi_handler);
 struct event_constraint emptyconstraint;
 struct event_constraint unconstrained;
 
-static int
-x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
+static int x86_pmu_prepare_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (long)hcpu;
 	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
-	int i, ret = NOTIFY_OK;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		for (i = 0 ; i < X86_PERF_KFREE_MAX; i++)
-			cpuc->kfree_on_online[i] = NULL;
-		if (x86_pmu.cpu_prepare)
-			ret = x86_pmu.cpu_prepare(cpu);
-		break;
-
-	case CPU_STARTING:
-		if (x86_pmu.cpu_starting)
-			x86_pmu.cpu_starting(cpu);
-		break;
+	int i;
 
-	case CPU_ONLINE:
-		for (i = 0 ; i < X86_PERF_KFREE_MAX; i++) {
-			kfree(cpuc->kfree_on_online[i]);
-			cpuc->kfree_on_online[i] = NULL;
-		}
-		break;
+	for (i = 0 ; i < X86_PERF_KFREE_MAX; i++)
+		cpuc->kfree_on_online[i] = NULL;
+	if (x86_pmu.cpu_prepare)
+		return x86_pmu.cpu_prepare(cpu);
+	return 0;
+}
 
-	case CPU_DYING:
-		if (x86_pmu.cpu_dying)
-			x86_pmu.cpu_dying(cpu);
-		break;
+static int x86_pmu_dead_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_dead)
+		x86_pmu.cpu_dead(cpu);
+	return 0;
+}
 
-	case CPU_UP_CANCELED:
-	case CPU_DEAD:
-		if (x86_pmu.cpu_dead)
-			x86_pmu.cpu_dead(cpu);
-		break;
+static int x86_pmu_online_cpu(unsigned int cpu)
+{
+	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
+	int i;
 
-	default:
-		break;
+	for (i = 0 ; i < X86_PERF_KFREE_MAX; i++) {
+		kfree(cpuc->kfree_on_online[i]);
+		cpuc->kfree_on_online[i] = NULL;
 	}
+	return 0;
+}
 
-	return ret;
+static int x86_pmu_starting_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_starting)
+		x86_pmu.cpu_starting(cpu);
+	return 0;
+}
+
+static int x86_pmu_dying_cpu(unsigned int cpu)
+{
+	if (x86_pmu.cpu_dying)
+		x86_pmu.cpu_dying(cpu);
+	return 0;
 }
 
 static void __init pmu_check_apic(void)
@@ -1764,10 +1764,39 @@ static int __init init_hw_perf_events(void)
 	pr_info("... fixed-purpose events:   %d\n",     x86_pmu.num_counters_fixed);
 	pr_info("... event mask:             %016Lx\n", x86_pmu.intel_ctrl);
 
-	perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(x86_pmu_notifier);
+	/*
+	 * Install callbacks. Core will call them for each online
+	 * cpu.
+	 */
+	err = cpuhp_setup_state(CPUHP_PERF_X86_PREPARE, "PERF_X86_PREPARE",
+				x86_pmu_prepare_cpu, x86_pmu_dead_cpu);
+	if (err)
+		return err;
+
+	err = cpuhp_setup_state(CPUHP_AP_PERF_X86_STARTING,
+				"AP_PERF_X86_STARTING", x86_pmu_starting_cpu,
+				x86_pmu_dying_cpu);
+	if (err)
+		goto out;
+
+	err = cpuhp_setup_state(CPUHP_AP_PERF_X86_ONLINE, "AP_PERF_X86_ONLINE",
+				x86_pmu_online_cpu, NULL);
+	if (err)
+		goto out1;
+
+	err = perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
+	if (err)
+		goto out2;
 
 	return 0;
+
+out2:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_ONLINE);
+out1:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_STARTING);
+out:
+	cpuhp_remove_state(CPUHP_PERF_X86_PREPARE);
+	return err;
 }
 early_initcall(init_hw_perf_events);
 
diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index 9b4f9d3..6a1441b 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -3050,7 +3050,7 @@ static int intel_pmu_cpu_prepare(int cpu)
 		cpuc->excl_thread_id = 0;
 	}
 
-	return NOTIFY_OK;
+	return 0;
 
 err_constraint_list:
 	kfree(cpuc->constraint_list);
@@ -3061,7 +3061,7 @@ err_shared_regs:
 	cpuc->shared_regs = NULL;
 
 err:
-	return NOTIFY_BAD;
+	return -ENOMEM;
 }
 
 static void intel_pmu_cpu_starting(int cpu)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 067082e..858915f 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -5,6 +5,7 @@ enum cpuhp_state {
 	CPUHP_OFFLINE,
 	CPUHP_CREATE_THREADS,
 	CPUHP_PERF_PREPARE,
+	CPUHP_PERF_X86_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -17,6 +18,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
+	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -24,6 +26,7 @@ enum cpuhp_state {
 	CPUHP_AP_SMPBOOT_THREADS,
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_PERF_ONLINE,
+	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 11/67] perf/x86/intel/uncore: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0010-perf-x86-intel-uncore-Convert-to-hotplug-state-machi.patch --]
[-- Type: text/plain, Size: 8379 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Convert the notifiers to state machine states and let the core code do the
setup for the already online CPUs. This notifier has a completely undocumented
ordering requirement versus perf hardcoded in the notifier priority. This
odering is only required for CPU down, so that hardware migration happens
before the core is notified about the outgoing CPU.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/intel/uncore.c | 133 ++++++++++++++---------------------------
 include/linux/cpuhotplug.h     |   3 +
 2 files changed, 48 insertions(+), 88 deletions(-)

diff --git a/arch/x86/events/intel/uncore.c b/arch/x86/events/intel/uncore.c
index fce7406..8e280a7 100644
--- a/arch/x86/events/intel/uncore.c
+++ b/arch/x86/events/intel/uncore.c
@@ -1034,7 +1034,7 @@ static void uncore_pci_exit(void)
 	}
 }
 
-static void uncore_cpu_dying(int cpu)
+static int uncore_cpu_dying(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
@@ -1051,16 +1051,19 @@ static void uncore_cpu_dying(int cpu)
 				uncore_box_exit(box);
 		}
 	}
+	return 0;
 }
 
-static void uncore_cpu_starting(int cpu, bool init)
+static int first_init;
+
+static int uncore_cpu_starting(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
 	struct intel_uncore_box *box;
 	int i, pkg, ncpus = 1;
 
-	if (init) {
+	if (first_init) {
 		/*
 		 * On init we get the number of online cpus in the package
 		 * and set refcount for all of them.
@@ -1081,9 +1084,11 @@ static void uncore_cpu_starting(int cpu, bool init)
 				uncore_box_init(box);
 		}
 	}
+
+	return 0;
 }
 
-static int uncore_cpu_prepare(int cpu)
+static int uncore_cpu_prepare(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
@@ -1146,13 +1151,13 @@ static void uncore_change_context(struct intel_uncore_type **uncores,
 		uncore_change_type_ctx(*uncores, old_cpu, new_cpu);
 }
 
-static void uncore_event_exit_cpu(int cpu)
+static int uncore_event_cpu_offline(unsigned int cpu)
 {
 	int target;
 
 	/* Check if exiting cpu is used for collecting uncore events */
 	if (!cpumask_test_and_clear_cpu(cpu, &uncore_cpu_mask))
-		return;
+		return 0;
 
 	/* Find a new cpu to collect uncore events */
 	target = cpumask_any_but(topology_core_cpumask(cpu), cpu);
@@ -1165,9 +1170,10 @@ static void uncore_event_exit_cpu(int cpu)
 
 	uncore_change_context(uncore_msr_uncores, cpu, target);
 	uncore_change_context(uncore_pci_uncores, cpu, target);
+	return 0;
 }
 
-static void uncore_event_init_cpu(int cpu)
+static int uncore_event_cpu_online(unsigned int cpu)
 {
 	int target;
 
@@ -1177,50 +1183,15 @@ static void uncore_event_init_cpu(int cpu)
 	 */
 	target = cpumask_any_and(&uncore_cpu_mask, topology_core_cpumask(cpu));
 	if (target < nr_cpu_ids)
-		return;
+		return 0;
 
 	cpumask_set_cpu(cpu, &uncore_cpu_mask);
 
 	uncore_change_context(uncore_msr_uncores, -1, cpu);
 	uncore_change_context(uncore_pci_uncores, -1, cpu);
+	return 0;
 }
 
-static int uncore_cpu_notifier(struct notifier_block *self,
-			       unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		return notifier_from_errno(uncore_cpu_prepare(cpu));
-
-	case CPU_STARTING:
-		uncore_cpu_starting(cpu, false);
-	case CPU_DOWN_FAILED:
-		uncore_event_init_cpu(cpu);
-		break;
-
-	case CPU_UP_CANCELED:
-	case CPU_DYING:
-		uncore_cpu_dying(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		uncore_event_exit_cpu(cpu);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block uncore_cpu_nb = {
-	.notifier_call	= uncore_cpu_notifier,
-	/*
-	 * to migrate uncore events, our notifier should be executed
-	 * before perf core's notifier.
-	 */
-	.priority	= CPU_PRI_PERF + 1,
-};
-
 static int __init type_pmu_register(struct intel_uncore_type *type)
 {
 	int i, ret;
@@ -1264,41 +1235,6 @@ err:
 	return ret;
 }
 
-static void __init uncore_cpu_setup(void *dummy)
-{
-	uncore_cpu_starting(smp_processor_id(), true);
-}
-
-/* Lazy to avoid allocation of a few bytes for the normal case */
-static __initdata DECLARE_BITMAP(packages, MAX_LOCAL_APIC);
-
-static int __init uncore_cpumask_init(bool msr)
-{
-	unsigned int cpu;
-
-	for_each_online_cpu(cpu) {
-		unsigned int pkg = topology_logical_package_id(cpu);
-		int ret;
-
-		if (test_and_set_bit(pkg, packages))
-			continue;
-		/*
-		 * The first online cpu of each package allocates and takes
-		 * the refcounts for all other online cpus in that package.
-		 * If msrs are not enabled no allocation is required.
-		 */
-		if (msr) {
-			ret = uncore_cpu_prepare(cpu);
-			if (ret)
-				return ret;
-		}
-		uncore_event_init_cpu(cpu);
-		smp_call_function_single(cpu, uncore_cpu_setup, NULL, 1);
-	}
-	__register_cpu_notifier(&uncore_cpu_nb);
-	return 0;
-}
-
 #define X86_UNCORE_MODEL_MATCH(model, init)	\
 	{ X86_VENDOR_INTEL, 6, model, X86_FEATURE_ANY, (unsigned long)&init }
 
@@ -1420,11 +1356,33 @@ static int __init intel_uncore_init(void)
 	if (cret && pret)
 		return -ENODEV;
 
-	cpu_notifier_register_begin();
-	ret = uncore_cpumask_init(!cret);
-	if (ret)
-		goto err;
-	cpu_notifier_register_done();
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 *
+	 * The first online cpu of each package allocates and takes
+	 * the refcounts for all other online cpus in that package.
+	 * If msrs are not enabled no allocation is required and
+	 * uncore_cpu_prepare() is not called for each online cpu.
+	 */
+	if (!cret) {
+	       ret = cpuhp_setup_state(CPUHP_PERF_X86_UNCORE_PREP,
+					"PERF_X86_UNCORE_PREP",
+					uncore_cpu_prepare, NULL);
+		if (ret)
+			goto err;
+	} else {
+		cpuhp_setup_state_nocalls(CPUHP_PERF_X86_UNCORE_PREP,
+					  "PERF_X86_UNCORE_PREP",
+					  uncore_cpu_prepare, NULL);
+	}
+	first_init = 1;
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_UNCORE_STARTING,
+			  "AP_PERF_X86_UNCORE_STARTING",
+			  uncore_cpu_starting, uncore_cpu_dying);
+	first_init = 0;
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_UNCORE_ONLINE,
+			  "AP_PERF_X86_UNCORE_ONLINE",
+			  uncore_event_cpu_online, uncore_event_cpu_offline);
 	return 0;
 
 err:
@@ -1432,17 +1390,16 @@ err:
 	on_each_cpu_mask(&uncore_cpu_mask, uncore_exit_boxes, NULL, 1);
 	uncore_types_exit(uncore_msr_uncores);
 	uncore_pci_exit();
-	cpu_notifier_register_done();
 	return ret;
 }
 module_init(intel_uncore_init);
 
 static void __exit intel_uncore_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&uncore_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_UNCORE_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_UNCORE_STARTING);
+	cpuhp_remove_state_nocalls(CPUHP_PERF_X86_UNCORE_PREP);
 	uncore_types_exit(uncore_msr_uncores);
 	uncore_pci_exit();
-	cpu_notifier_register_done();
 }
 module_exit(intel_uncore_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 858915f..6f91e94 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -6,6 +6,7 @@ enum cpuhp_state {
 	CPUHP_CREATE_THREADS,
 	CPUHP_PERF_PREPARE,
 	CPUHP_PERF_X86_PREPARE,
+	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -18,6 +19,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
+	CPUHP_AP_PERF_X86_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
@@ -27,6 +29,7 @@ enum cpuhp_state {
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_PERF_X86_ONLINE,
+	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/intel/uncore: Convert to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  1a246b9f58c6149b5a5bec081418b8ed890e0dfe
Gitweb:     http://git.kernel.org/tip/1a246b9f58c6149b5a5bec081418b8ed890e0dfe
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:12 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:32 +0200

perf/x86/intel/uncore: Convert to hotplug state machine

Convert the notifiers to state machine states and let the core code do the
setup for the already online CPUs. This notifier has a completely undocumented
ordering requirement versus perf hardcoded in the notifier priority. This
odering is only required for CPU down, so that hardware migration happens
before the core is notified about the outgoing CPU.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.752695801@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/intel/uncore.c | 133 ++++++++++++++---------------------------
 include/linux/cpuhotplug.h     |   3 +
 2 files changed, 48 insertions(+), 88 deletions(-)

diff --git a/arch/x86/events/intel/uncore.c b/arch/x86/events/intel/uncore.c
index fce7406..8e280a7 100644
--- a/arch/x86/events/intel/uncore.c
+++ b/arch/x86/events/intel/uncore.c
@@ -1034,7 +1034,7 @@ static void uncore_pci_exit(void)
 	}
 }
 
-static void uncore_cpu_dying(int cpu)
+static int uncore_cpu_dying(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
@@ -1051,16 +1051,19 @@ static void uncore_cpu_dying(int cpu)
 				uncore_box_exit(box);
 		}
 	}
+	return 0;
 }
 
-static void uncore_cpu_starting(int cpu, bool init)
+static int first_init;
+
+static int uncore_cpu_starting(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
 	struct intel_uncore_box *box;
 	int i, pkg, ncpus = 1;
 
-	if (init) {
+	if (first_init) {
 		/*
 		 * On init we get the number of online cpus in the package
 		 * and set refcount for all of them.
@@ -1081,9 +1084,11 @@ static void uncore_cpu_starting(int cpu, bool init)
 				uncore_box_init(box);
 		}
 	}
+
+	return 0;
 }
 
-static int uncore_cpu_prepare(int cpu)
+static int uncore_cpu_prepare(unsigned int cpu)
 {
 	struct intel_uncore_type *type, **types = uncore_msr_uncores;
 	struct intel_uncore_pmu *pmu;
@@ -1146,13 +1151,13 @@ static void uncore_change_context(struct intel_uncore_type **uncores,
 		uncore_change_type_ctx(*uncores, old_cpu, new_cpu);
 }
 
-static void uncore_event_exit_cpu(int cpu)
+static int uncore_event_cpu_offline(unsigned int cpu)
 {
 	int target;
 
 	/* Check if exiting cpu is used for collecting uncore events */
 	if (!cpumask_test_and_clear_cpu(cpu, &uncore_cpu_mask))
-		return;
+		return 0;
 
 	/* Find a new cpu to collect uncore events */
 	target = cpumask_any_but(topology_core_cpumask(cpu), cpu);
@@ -1165,9 +1170,10 @@ static void uncore_event_exit_cpu(int cpu)
 
 	uncore_change_context(uncore_msr_uncores, cpu, target);
 	uncore_change_context(uncore_pci_uncores, cpu, target);
+	return 0;
 }
 
-static void uncore_event_init_cpu(int cpu)
+static int uncore_event_cpu_online(unsigned int cpu)
 {
 	int target;
 
@@ -1177,50 +1183,15 @@ static void uncore_event_init_cpu(int cpu)
 	 */
 	target = cpumask_any_and(&uncore_cpu_mask, topology_core_cpumask(cpu));
 	if (target < nr_cpu_ids)
-		return;
+		return 0;
 
 	cpumask_set_cpu(cpu, &uncore_cpu_mask);
 
 	uncore_change_context(uncore_msr_uncores, -1, cpu);
 	uncore_change_context(uncore_pci_uncores, -1, cpu);
+	return 0;
 }
 
-static int uncore_cpu_notifier(struct notifier_block *self,
-			       unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		return notifier_from_errno(uncore_cpu_prepare(cpu));
-
-	case CPU_STARTING:
-		uncore_cpu_starting(cpu, false);
-	case CPU_DOWN_FAILED:
-		uncore_event_init_cpu(cpu);
-		break;
-
-	case CPU_UP_CANCELED:
-	case CPU_DYING:
-		uncore_cpu_dying(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		uncore_event_exit_cpu(cpu);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block uncore_cpu_nb = {
-	.notifier_call	= uncore_cpu_notifier,
-	/*
-	 * to migrate uncore events, our notifier should be executed
-	 * before perf core's notifier.
-	 */
-	.priority	= CPU_PRI_PERF + 1,
-};
-
 static int __init type_pmu_register(struct intel_uncore_type *type)
 {
 	int i, ret;
@@ -1264,41 +1235,6 @@ err:
 	return ret;
 }
 
-static void __init uncore_cpu_setup(void *dummy)
-{
-	uncore_cpu_starting(smp_processor_id(), true);
-}
-
-/* Lazy to avoid allocation of a few bytes for the normal case */
-static __initdata DECLARE_BITMAP(packages, MAX_LOCAL_APIC);
-
-static int __init uncore_cpumask_init(bool msr)
-{
-	unsigned int cpu;
-
-	for_each_online_cpu(cpu) {
-		unsigned int pkg = topology_logical_package_id(cpu);
-		int ret;
-
-		if (test_and_set_bit(pkg, packages))
-			continue;
-		/*
-		 * The first online cpu of each package allocates and takes
-		 * the refcounts for all other online cpus in that package.
-		 * If msrs are not enabled no allocation is required.
-		 */
-		if (msr) {
-			ret = uncore_cpu_prepare(cpu);
-			if (ret)
-				return ret;
-		}
-		uncore_event_init_cpu(cpu);
-		smp_call_function_single(cpu, uncore_cpu_setup, NULL, 1);
-	}
-	__register_cpu_notifier(&uncore_cpu_nb);
-	return 0;
-}
-
 #define X86_UNCORE_MODEL_MATCH(model, init)	\
 	{ X86_VENDOR_INTEL, 6, model, X86_FEATURE_ANY, (unsigned long)&init }
 
@@ -1420,11 +1356,33 @@ static int __init intel_uncore_init(void)
 	if (cret && pret)
 		return -ENODEV;
 
-	cpu_notifier_register_begin();
-	ret = uncore_cpumask_init(!cret);
-	if (ret)
-		goto err;
-	cpu_notifier_register_done();
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 *
+	 * The first online cpu of each package allocates and takes
+	 * the refcounts for all other online cpus in that package.
+	 * If msrs are not enabled no allocation is required and
+	 * uncore_cpu_prepare() is not called for each online cpu.
+	 */
+	if (!cret) {
+	       ret = cpuhp_setup_state(CPUHP_PERF_X86_UNCORE_PREP,
+					"PERF_X86_UNCORE_PREP",
+					uncore_cpu_prepare, NULL);
+		if (ret)
+			goto err;
+	} else {
+		cpuhp_setup_state_nocalls(CPUHP_PERF_X86_UNCORE_PREP,
+					  "PERF_X86_UNCORE_PREP",
+					  uncore_cpu_prepare, NULL);
+	}
+	first_init = 1;
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_UNCORE_STARTING,
+			  "AP_PERF_X86_UNCORE_STARTING",
+			  uncore_cpu_starting, uncore_cpu_dying);
+	first_init = 0;
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_UNCORE_ONLINE,
+			  "AP_PERF_X86_UNCORE_ONLINE",
+			  uncore_event_cpu_online, uncore_event_cpu_offline);
 	return 0;
 
 err:
@@ -1432,17 +1390,16 @@ err:
 	on_each_cpu_mask(&uncore_cpu_mask, uncore_exit_boxes, NULL, 1);
 	uncore_types_exit(uncore_msr_uncores);
 	uncore_pci_exit();
-	cpu_notifier_register_done();
 	return ret;
 }
 module_init(intel_uncore_init);
 
 static void __exit intel_uncore_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&uncore_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_UNCORE_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_UNCORE_STARTING);
+	cpuhp_remove_state_nocalls(CPUHP_PERF_X86_UNCORE_PREP);
 	uncore_types_exit(uncore_msr_uncores);
 	uncore_pci_exit();
-	cpu_notifier_register_done();
 }
 module_exit(intel_uncore_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 858915f..6f91e94 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -6,6 +6,7 @@ enum cpuhp_state {
 	CPUHP_CREATE_THREADS,
 	CPUHP_PERF_PREPARE,
 	CPUHP_PERF_X86_PREPARE,
+	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -18,6 +19,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_ARMADA_CASC_STARTING,
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
+	CPUHP_AP_PERF_X86_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
@@ -27,6 +29,7 @@ enum cpuhp_state {
 	CPUHP_AP_X86_VDSO_VMA_ONLINE,
 	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_PERF_X86_ONLINE,
+	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 12/67] perf/x86/amd/uncore: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0011-perf-x86-amd-uncore-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 6726 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chen Yucong <slaoub@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/amd/uncore.c | 122 ++++++++++++-------------------------------
 include/linux/cpuhotplug.h   |   3 ++
 2 files changed, 35 insertions(+), 90 deletions(-)

diff --git a/arch/x86/events/amd/uncore.c b/arch/x86/events/amd/uncore.c
index 98ac573..e6131d4 100644
--- a/arch/x86/events/amd/uncore.c
+++ b/arch/x86/events/amd/uncore.c
@@ -358,7 +358,7 @@ amd_uncore_find_online_sibling(struct amd_uncore *this,
 	return this;
 }
 
-static void amd_uncore_cpu_starting(unsigned int cpu)
+static int amd_uncore_cpu_starting(unsigned int cpu)
 {
 	unsigned int eax, ebx, ecx, edx;
 	struct amd_uncore *uncore;
@@ -384,6 +384,8 @@ static void amd_uncore_cpu_starting(unsigned int cpu)
 		uncore = amd_uncore_find_online_sibling(uncore, amd_uncore_l2);
 		*per_cpu_ptr(amd_uncore_l2, cpu) = uncore;
 	}
+
+	return 0;
 }
 
 static void uncore_online(unsigned int cpu,
@@ -398,13 +400,15 @@ static void uncore_online(unsigned int cpu,
 		cpumask_set_cpu(cpu, uncore->active_mask);
 }
 
-static void amd_uncore_cpu_online(unsigned int cpu)
+static int amd_uncore_cpu_online(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_online(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_online(cpu, amd_uncore_l2);
+
+	return 0;
 }
 
 static void uncore_down_prepare(unsigned int cpu,
@@ -433,13 +437,15 @@ static void uncore_down_prepare(unsigned int cpu,
 	}
 }
 
-static void amd_uncore_cpu_down_prepare(unsigned int cpu)
+static int amd_uncore_cpu_down_prepare(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_down_prepare(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_down_prepare(cpu, amd_uncore_l2);
+
+	return 0;
 }
 
 static void uncore_dead(unsigned int cpu, struct amd_uncore * __percpu *uncores)
@@ -454,74 +460,19 @@ static void uncore_dead(unsigned int cpu, struct amd_uncore * __percpu *uncores)
 	*per_cpu_ptr(uncores, cpu) = NULL;
 }
 
-static void amd_uncore_cpu_dead(unsigned int cpu)
+static int amd_uncore_cpu_dead(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_dead(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_dead(cpu, amd_uncore_l2);
-}
-
-static int
-amd_uncore_cpu_notifier(struct notifier_block *self, unsigned long action,
-			void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		if (amd_uncore_cpu_up_prepare(cpu))
-			return notifier_from_errno(-ENOMEM);
-		break;
-
-	case CPU_STARTING:
-		amd_uncore_cpu_starting(cpu);
-		break;
-
-	case CPU_ONLINE:
-		amd_uncore_cpu_online(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		amd_uncore_cpu_down_prepare(cpu);
-		break;
-
-	case CPU_UP_CANCELED:
-	case CPU_DEAD:
-		amd_uncore_cpu_dead(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block amd_uncore_cpu_notifier_block = {
-	.notifier_call	= amd_uncore_cpu_notifier,
-	.priority	= CPU_PRI_PERF + 1,
-};
-
-static void __init init_cpu_already_online(void *dummy)
-{
-	unsigned int cpu = smp_processor_id();
-
-	amd_uncore_cpu_starting(cpu);
-	amd_uncore_cpu_online(cpu);
-}
 
-static void cleanup_cpu_online(void *dummy)
-{
-	unsigned int cpu = smp_processor_id();
-
-	amd_uncore_cpu_dead(cpu);
+	return 0;
 }
 
 static int __init amd_uncore_init(void)
 {
-	unsigned int cpu, cpu2;
 	int ret = -ENODEV;
 
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
@@ -558,38 +509,29 @@ static int __init amd_uncore_init(void)
 		ret = 0;
 	}
 
-	if (ret)
-		goto fail_nodev;
-
-	cpu_notifier_register_begin();
-
-	/* init cpus already online before registering for hotplug notifier */
-	for_each_online_cpu(cpu) {
-		ret = amd_uncore_cpu_up_prepare(cpu);
-		if (ret)
-			goto fail_online;
-		smp_call_function_single(cpu, init_cpu_already_online, NULL, 1);
-	}
-
-	__register_cpu_notifier(&amd_uncore_cpu_notifier_block);
-	cpu_notifier_register_done();
-
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 */
+	if (cpuhp_setup_state(CPUHP_PERF_X86_AMD_UNCORE_PREP,
+			      "PERF_X86_AMD_UNCORE_PREP",
+			      amd_uncore_cpu_up_prepare, amd_uncore_cpu_dead))
+		goto fail_l2;
+
+	if (cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
+			      "AP_PERF_X86_AMD_UNCORE_STARTING",
+			      amd_uncore_cpu_starting, NULL))
+		goto fail_prep;
+	if (cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+			      "AP_PERF_X86_AMD_UNCORE_ONLINE",
+			      amd_uncore_cpu_online,
+			      amd_uncore_cpu_down_prepare))
+		goto fail_start;
 	return 0;
 
-
-fail_online:
-	for_each_online_cpu(cpu2) {
-		if (cpu2 == cpu)
-			break;
-		smp_call_function_single(cpu, cleanup_cpu_online, NULL, 1);
-	}
-	cpu_notifier_register_done();
-
-	/* amd_uncore_nb/l2 should have been freed by cleanup_cpu_online */
-	amd_uncore_nb = amd_uncore_l2 = NULL;
-
-	if (boot_cpu_has(X86_FEATURE_PERFCTR_L2))
-		perf_pmu_unregister(&amd_l2_pmu);
+fail_start:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING);
+fail_prep:
+	cpuhp_remove_state(CPUHP_PERF_X86_AMD_UNCORE_PREP);
 fail_l2:
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_NB))
 		perf_pmu_unregister(&amd_nb_pmu);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6f91e94..569d057 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -7,6 +7,7 @@ enum cpuhp_state {
 	CPUHP_PERF_PREPARE,
 	CPUHP_PERF_X86_PREPARE,
 	CPUHP_PERF_X86_UNCORE_PREP,
+	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -20,6 +21,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
 	CPUHP_AP_PERF_X86_UNCORE_STARTING,
+	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
@@ -30,6 +32,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/amd/uncore: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  96b2bd3866a0b045330e420a2f1829ff2a3399bc
Gitweb:     http://git.kernel.org/tip/96b2bd3866a0b045330e420a2f1829ff2a3399bc
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:13 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:33 +0200

perf/x86/amd/uncore: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chen Yucong <slaoub@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.839150380@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/amd/uncore.c | 122 ++++++++++++-------------------------------
 include/linux/cpuhotplug.h   |   3 ++
 2 files changed, 35 insertions(+), 90 deletions(-)

diff --git a/arch/x86/events/amd/uncore.c b/arch/x86/events/amd/uncore.c
index 98ac573..e6131d4 100644
--- a/arch/x86/events/amd/uncore.c
+++ b/arch/x86/events/amd/uncore.c
@@ -358,7 +358,7 @@ amd_uncore_find_online_sibling(struct amd_uncore *this,
 	return this;
 }
 
-static void amd_uncore_cpu_starting(unsigned int cpu)
+static int amd_uncore_cpu_starting(unsigned int cpu)
 {
 	unsigned int eax, ebx, ecx, edx;
 	struct amd_uncore *uncore;
@@ -384,6 +384,8 @@ static void amd_uncore_cpu_starting(unsigned int cpu)
 		uncore = amd_uncore_find_online_sibling(uncore, amd_uncore_l2);
 		*per_cpu_ptr(amd_uncore_l2, cpu) = uncore;
 	}
+
+	return 0;
 }
 
 static void uncore_online(unsigned int cpu,
@@ -398,13 +400,15 @@ static void uncore_online(unsigned int cpu,
 		cpumask_set_cpu(cpu, uncore->active_mask);
 }
 
-static void amd_uncore_cpu_online(unsigned int cpu)
+static int amd_uncore_cpu_online(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_online(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_online(cpu, amd_uncore_l2);
+
+	return 0;
 }
 
 static void uncore_down_prepare(unsigned int cpu,
@@ -433,13 +437,15 @@ static void uncore_down_prepare(unsigned int cpu,
 	}
 }
 
-static void amd_uncore_cpu_down_prepare(unsigned int cpu)
+static int amd_uncore_cpu_down_prepare(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_down_prepare(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_down_prepare(cpu, amd_uncore_l2);
+
+	return 0;
 }
 
 static void uncore_dead(unsigned int cpu, struct amd_uncore * __percpu *uncores)
@@ -454,74 +460,19 @@ static void uncore_dead(unsigned int cpu, struct amd_uncore * __percpu *uncores)
 	*per_cpu_ptr(uncores, cpu) = NULL;
 }
 
-static void amd_uncore_cpu_dead(unsigned int cpu)
+static int amd_uncore_cpu_dead(unsigned int cpu)
 {
 	if (amd_uncore_nb)
 		uncore_dead(cpu, amd_uncore_nb);
 
 	if (amd_uncore_l2)
 		uncore_dead(cpu, amd_uncore_l2);
-}
-
-static int
-amd_uncore_cpu_notifier(struct notifier_block *self, unsigned long action,
-			void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		if (amd_uncore_cpu_up_prepare(cpu))
-			return notifier_from_errno(-ENOMEM);
-		break;
-
-	case CPU_STARTING:
-		amd_uncore_cpu_starting(cpu);
-		break;
-
-	case CPU_ONLINE:
-		amd_uncore_cpu_online(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		amd_uncore_cpu_down_prepare(cpu);
-		break;
-
-	case CPU_UP_CANCELED:
-	case CPU_DEAD:
-		amd_uncore_cpu_dead(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block amd_uncore_cpu_notifier_block = {
-	.notifier_call	= amd_uncore_cpu_notifier,
-	.priority	= CPU_PRI_PERF + 1,
-};
-
-static void __init init_cpu_already_online(void *dummy)
-{
-	unsigned int cpu = smp_processor_id();
-
-	amd_uncore_cpu_starting(cpu);
-	amd_uncore_cpu_online(cpu);
-}
 
-static void cleanup_cpu_online(void *dummy)
-{
-	unsigned int cpu = smp_processor_id();
-
-	amd_uncore_cpu_dead(cpu);
+	return 0;
 }
 
 static int __init amd_uncore_init(void)
 {
-	unsigned int cpu, cpu2;
 	int ret = -ENODEV;
 
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
@@ -558,38 +509,29 @@ static int __init amd_uncore_init(void)
 		ret = 0;
 	}
 
-	if (ret)
-		goto fail_nodev;
-
-	cpu_notifier_register_begin();
-
-	/* init cpus already online before registering for hotplug notifier */
-	for_each_online_cpu(cpu) {
-		ret = amd_uncore_cpu_up_prepare(cpu);
-		if (ret)
-			goto fail_online;
-		smp_call_function_single(cpu, init_cpu_already_online, NULL, 1);
-	}
-
-	__register_cpu_notifier(&amd_uncore_cpu_notifier_block);
-	cpu_notifier_register_done();
-
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 */
+	if (cpuhp_setup_state(CPUHP_PERF_X86_AMD_UNCORE_PREP,
+			      "PERF_X86_AMD_UNCORE_PREP",
+			      amd_uncore_cpu_up_prepare, amd_uncore_cpu_dead))
+		goto fail_l2;
+
+	if (cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
+			      "AP_PERF_X86_AMD_UNCORE_STARTING",
+			      amd_uncore_cpu_starting, NULL))
+		goto fail_prep;
+	if (cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+			      "AP_PERF_X86_AMD_UNCORE_ONLINE",
+			      amd_uncore_cpu_online,
+			      amd_uncore_cpu_down_prepare))
+		goto fail_start;
 	return 0;
 
-
-fail_online:
-	for_each_online_cpu(cpu2) {
-		if (cpu2 == cpu)
-			break;
-		smp_call_function_single(cpu, cleanup_cpu_online, NULL, 1);
-	}
-	cpu_notifier_register_done();
-
-	/* amd_uncore_nb/l2 should have been freed by cleanup_cpu_online */
-	amd_uncore_nb = amd_uncore_l2 = NULL;
-
-	if (boot_cpu_has(X86_FEATURE_PERFCTR_L2))
-		perf_pmu_unregister(&amd_l2_pmu);
+fail_start:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING);
+fail_prep:
+	cpuhp_remove_state(CPUHP_PERF_X86_AMD_UNCORE_PREP);
 fail_l2:
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_NB))
 		perf_pmu_unregister(&amd_nb_pmu);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6f91e94..569d057 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -7,6 +7,7 @@ enum cpuhp_state {
 	CPUHP_PERF_PREPARE,
 	CPUHP_PERF_X86_PREPARE,
 	CPUHP_PERF_X86_UNCORE_PREP,
+	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -20,6 +21,7 @@ enum cpuhp_state {
 	CPUHP_AP_IRQ_BCM2836_STARTING,
 	CPUHP_AP_ARM_MVEBU_COHERENCY,
 	CPUHP_AP_PERF_X86_UNCORE_STARTING,
+	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
@@ -30,6 +32,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ONLINE,
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 13/67] perf/x86/amd/ibs: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0012-perf-x86-amd-ibs-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 4640 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chen Yucong <slaoub@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/amd/ibs.c  | 64 ++++++++++++++++++++--------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 29 insertions(+), 36 deletions(-)

diff --git a/arch/x86/events/amd/ibs.c b/arch/x86/events/amd/ibs.c
index feb90f6..1a59a18 100644
--- a/arch/x86/events/amd/ibs.c
+++ b/arch/x86/events/amd/ibs.c
@@ -721,13 +721,10 @@ static __init int perf_ibs_pmu_init(struct perf_ibs *perf_ibs, char *name)
 	return ret;
 }
 
-static __init int perf_event_ibs_init(void)
+static __init void perf_event_ibs_init(void)
 {
 	struct attribute **attr = ibs_op_format_attrs;
 
-	if (!ibs_caps)
-		return -ENODEV;	/* ibs not supported by the cpu */
-
 	perf_ibs_pmu_init(&perf_ibs_fetch, "ibs_fetch");
 
 	if (ibs_caps & IBS_CAPS_OPCNT) {
@@ -738,13 +735,11 @@ static __init int perf_event_ibs_init(void)
 
 	register_nmi_handler(NMI_LOCAL, perf_ibs_nmi_handler, 0, "perf_ibs");
 	pr_info("perf: AMD IBS detected (0x%08x)\n", ibs_caps);
-
-	return 0;
 }
 
 #else /* defined(CONFIG_PERF_EVENTS) && defined(CONFIG_CPU_SUP_AMD) */
 
-static __init int perf_event_ibs_init(void) { return 0; }
+static __init void perf_event_ibs_init(void) { }
 
 #endif
 
@@ -921,7 +916,7 @@ static inline int get_ibs_lvt_offset(void)
 	return val & IBSCTL_LVT_OFFSET_MASK;
 }
 
-static void setup_APIC_ibs(void *dummy)
+static void setup_APIC_ibs(void)
 {
 	int offset;
 
@@ -936,7 +931,7 @@ failed:
 		smp_processor_id());
 }
 
-static void clear_APIC_ibs(void *dummy)
+static void clear_APIC_ibs(void)
 {
 	int offset;
 
@@ -945,18 +940,24 @@ static void clear_APIC_ibs(void *dummy)
 		setup_APIC_eilvt(offset, 0, APIC_EILVT_MSG_FIX, 1);
 }
 
+static int x86_pmu_amd_ibs_starting_cpu(unsigned int cpu)
+{
+	setup_APIC_ibs();
+	return 0;
+}
+
 #ifdef CONFIG_PM
 
 static int perf_ibs_suspend(void)
 {
-	clear_APIC_ibs(NULL);
+	clear_APIC_ibs();
 	return 0;
 }
 
 static void perf_ibs_resume(void)
 {
 	ibs_eilvt_setup();
-	setup_APIC_ibs(NULL);
+	setup_APIC_ibs();
 }
 
 static struct syscore_ops perf_ibs_syscore_ops = {
@@ -975,27 +976,15 @@ static inline void perf_ibs_pm_init(void) { }
 
 #endif
 
-static int
-perf_ibs_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
+static int x86_pmu_amd_ibs_dying_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		setup_APIC_ibs(NULL);
-		break;
-	case CPU_DYING:
-		clear_APIC_ibs(NULL);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	clear_APIC_ibs();
+	return 0;
 }
 
 static __init int amd_ibs_init(void)
 {
 	u32 caps;
-	int ret = -EINVAL;
 
 	caps = __get_ibs_caps();
 	if (!caps)
@@ -1004,22 +993,25 @@ static __init int amd_ibs_init(void)
 	ibs_eilvt_setup();
 
 	if (!ibs_eilvt_valid())
-		goto out;
+		return -EINVAL;
 
 	perf_ibs_pm_init();
-	cpu_notifier_register_begin();
+
 	ibs_caps = caps;
 	/* make ibs_caps visible to other cpus: */
 	smp_mb();
-	smp_call_function(setup_APIC_ibs, NULL, 1);
-	__perf_cpu_notifier(perf_ibs_cpu_notifier);
-	cpu_notifier_register_done();
+	/*
+	 * x86_pmu_amd_ibs_starting_cpu will be called from core on
+	 * all online cpus.
+	 */
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
+			  "AP_PERF_X86_AMD_IBS_STARTING",
+			  x86_pmu_amd_ibs_starting_cpu,
+			  x86_pmu_amd_ibs_dying_cpu);
 
-	ret = perf_event_ibs_init();
-out:
-	if (ret)
-		pr_err("Failed to setup IBS, %d\n", ret);
-	return ret;
+	perf_event_ibs_init();
+
+	return 0;
 }
 
 /* Since we need the pci subsystem to init ibs we can't do this earlier: */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 569d057..4672019 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -23,6 +23,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
+	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/amd/ibs: Convert to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  9744f7b7b37e78d7f18050856e626aea33058787
Gitweb:     http://git.kernel.org/tip/9744f7b7b37e78d7f18050856e626aea33058787
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:14 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:34 +0200

perf/x86/amd/ibs: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chen Yucong <slaoub@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153333.921401190@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/amd/ibs.c  | 64 ++++++++++++++++++++--------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 29 insertions(+), 36 deletions(-)

diff --git a/arch/x86/events/amd/ibs.c b/arch/x86/events/amd/ibs.c
index feb90f6..1a59a18 100644
--- a/arch/x86/events/amd/ibs.c
+++ b/arch/x86/events/amd/ibs.c
@@ -721,13 +721,10 @@ static __init int perf_ibs_pmu_init(struct perf_ibs *perf_ibs, char *name)
 	return ret;
 }
 
-static __init int perf_event_ibs_init(void)
+static __init void perf_event_ibs_init(void)
 {
 	struct attribute **attr = ibs_op_format_attrs;
 
-	if (!ibs_caps)
-		return -ENODEV;	/* ibs not supported by the cpu */
-
 	perf_ibs_pmu_init(&perf_ibs_fetch, "ibs_fetch");
 
 	if (ibs_caps & IBS_CAPS_OPCNT) {
@@ -738,13 +735,11 @@ static __init int perf_event_ibs_init(void)
 
 	register_nmi_handler(NMI_LOCAL, perf_ibs_nmi_handler, 0, "perf_ibs");
 	pr_info("perf: AMD IBS detected (0x%08x)\n", ibs_caps);
-
-	return 0;
 }
 
 #else /* defined(CONFIG_PERF_EVENTS) && defined(CONFIG_CPU_SUP_AMD) */
 
-static __init int perf_event_ibs_init(void) { return 0; }
+static __init void perf_event_ibs_init(void) { }
 
 #endif
 
@@ -921,7 +916,7 @@ static inline int get_ibs_lvt_offset(void)
 	return val & IBSCTL_LVT_OFFSET_MASK;
 }
 
-static void setup_APIC_ibs(void *dummy)
+static void setup_APIC_ibs(void)
 {
 	int offset;
 
@@ -936,7 +931,7 @@ failed:
 		smp_processor_id());
 }
 
-static void clear_APIC_ibs(void *dummy)
+static void clear_APIC_ibs(void)
 {
 	int offset;
 
@@ -945,18 +940,24 @@ static void clear_APIC_ibs(void *dummy)
 		setup_APIC_eilvt(offset, 0, APIC_EILVT_MSG_FIX, 1);
 }
 
+static int x86_pmu_amd_ibs_starting_cpu(unsigned int cpu)
+{
+	setup_APIC_ibs();
+	return 0;
+}
+
 #ifdef CONFIG_PM
 
 static int perf_ibs_suspend(void)
 {
-	clear_APIC_ibs(NULL);
+	clear_APIC_ibs();
 	return 0;
 }
 
 static void perf_ibs_resume(void)
 {
 	ibs_eilvt_setup();
-	setup_APIC_ibs(NULL);
+	setup_APIC_ibs();
 }
 
 static struct syscore_ops perf_ibs_syscore_ops = {
@@ -975,27 +976,15 @@ static inline void perf_ibs_pm_init(void) { }
 
 #endif
 
-static int
-perf_ibs_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
+static int x86_pmu_amd_ibs_dying_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		setup_APIC_ibs(NULL);
-		break;
-	case CPU_DYING:
-		clear_APIC_ibs(NULL);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	clear_APIC_ibs();
+	return 0;
 }
 
 static __init int amd_ibs_init(void)
 {
 	u32 caps;
-	int ret = -EINVAL;
 
 	caps = __get_ibs_caps();
 	if (!caps)
@@ -1004,22 +993,25 @@ static __init int amd_ibs_init(void)
 	ibs_eilvt_setup();
 
 	if (!ibs_eilvt_valid())
-		goto out;
+		return -EINVAL;
 
 	perf_ibs_pm_init();
-	cpu_notifier_register_begin();
+
 	ibs_caps = caps;
 	/* make ibs_caps visible to other cpus: */
 	smp_mb();
-	smp_call_function(setup_APIC_ibs, NULL, 1);
-	__perf_cpu_notifier(perf_ibs_cpu_notifier);
-	cpu_notifier_register_done();
+	/*
+	 * x86_pmu_amd_ibs_starting_cpu will be called from core on
+	 * all online cpus.
+	 */
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
+			  "AP_PERF_X86_AMD_IBS_STARTING",
+			  x86_pmu_amd_ibs_starting_cpu,
+			  x86_pmu_amd_ibs_dying_cpu);
 
-	ret = perf_event_ibs_init();
-out:
-	if (ret)
-		pr_err("Failed to setup IBS, %d\n", ret);
-	return ret;
+	perf_event_ibs_init();
+
+	return 0;
 }
 
 /* Since we need the pci subsystem to init ibs we can't do this earlier: */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 569d057..4672019 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -23,6 +23,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
+	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 14/67] perf/x86/intel/rapl: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0013-perf-x86-intel-rapl-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 5525 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/intel/rapl.c | 84 +++++++++++++-------------------------------
 include/linux/cpuhotplug.h   |  2 ++
 2 files changed, 27 insertions(+), 59 deletions(-)

diff --git a/arch/x86/events/intel/rapl.c b/arch/x86/events/intel/rapl.c
index e30eef4..6255ede 100644
--- a/arch/x86/events/intel/rapl.c
+++ b/arch/x86/events/intel/rapl.c
@@ -555,14 +555,14 @@ const struct attribute_group *rapl_attr_groups[] = {
 	NULL,
 };
 
-static void rapl_cpu_exit(int cpu)
+static int rapl_cpu_offline(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
 
 	/* Check if exiting cpu is used for collecting rapl events */
 	if (!cpumask_test_and_clear_cpu(cpu, &rapl_cpu_mask))
-		return;
+		return 0;
 
 	pmu->cpu = -1;
 	/* Find a new cpu to collect rapl events */
@@ -574,9 +574,10 @@ static void rapl_cpu_exit(int cpu)
 		pmu->cpu = target;
 		perf_pmu_migrate_context(pmu->pmu, cpu, target);
 	}
+	return 0;
 }
 
-static void rapl_cpu_init(int cpu)
+static int rapl_cpu_online(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
@@ -587,13 +588,14 @@ static void rapl_cpu_init(int cpu)
 	 */
 	target = cpumask_any_and(&rapl_cpu_mask, topology_core_cpumask(cpu));
 	if (target < nr_cpu_ids)
-		return;
+		return 0;
 
 	cpumask_set_cpu(cpu, &rapl_cpu_mask);
 	pmu->cpu = cpu;
+	return 0;
 }
 
-static int rapl_cpu_prepare(int cpu)
+static int rapl_cpu_prepare(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 
@@ -614,33 +616,6 @@ static int rapl_cpu_prepare(int cpu)
 	return 0;
 }
 
-static int rapl_cpu_notifier(struct notifier_block *self,
-			     unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		rapl_cpu_prepare(cpu);
-		break;
-
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		rapl_cpu_init(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		rapl_cpu_exit(cpu);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block rapl_cpu_nb = {
-	.notifier_call	= rapl_cpu_notifier,
-	.priority       = CPU_PRI_PERF + 1,
-};
-
 static int rapl_check_hw_unit(bool apply_quirk)
 {
 	u64 msr_rapl_power_unit_bits;
@@ -691,24 +666,6 @@ static void __init rapl_advertise(void)
 	}
 }
 
-static int __init rapl_prepare_cpus(void)
-{
-	unsigned int cpu, pkg;
-	int ret;
-
-	for_each_online_cpu(cpu) {
-		pkg = topology_logical_package_id(cpu);
-		if (rapl_pmus->pmus[pkg])
-			continue;
-
-		ret = rapl_cpu_prepare(cpu);
-		if (ret)
-			return ret;
-		rapl_cpu_init(cpu);
-	}
-	return 0;
-}
-
 static void cleanup_rapl_pmus(void)
 {
 	int i;
@@ -835,35 +792,44 @@ static int __init rapl_pmu_init(void)
 	if (ret)
 		return ret;
 
-	cpu_notifier_register_begin();
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 */
 
-	ret = rapl_prepare_cpus();
+	ret = cpuhp_setup_state(CPUHP_PERF_X86_RAPL_PREP, "PERF_X86_RAPL_PREP",
+				rapl_cpu_prepare, NULL);
 	if (ret)
 		goto out;
 
+	ret = cpuhp_setup_state(CPUHP_AP_PERF_X86_RAPL_ONLINE,
+				"AP_PERF_X86_RAPL_ONLINE",
+				rapl_cpu_online, rapl_cpu_offline);
+	if (ret)
+		goto out1;
+
 	ret = perf_pmu_register(&rapl_pmus->pmu, "power", -1);
 	if (ret)
-		goto out;
+		goto out2;
 
-	__register_cpu_notifier(&rapl_cpu_nb);
-	cpu_notifier_register_done();
 	rapl_advertise();
 	return 0;
 
+out2:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_RAPL_ONLINE);
+out1:
+	cpuhp_remove_state(CPUHP_PERF_X86_RAPL_PREP);
 out:
 	pr_warn("Initialization failed (%d), disabled\n", ret);
 	cleanup_rapl_pmus();
-	cpu_notifier_register_done();
 	return ret;
 }
 module_init(rapl_pmu_init);
 
 static void __exit intel_rapl_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&rapl_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_RAPL_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_PERF_X86_RAPL_PREP);
 	perf_pmu_unregister(&rapl_pmus->pmu);
 	cleanup_rapl_pmus();
-	cpu_notifier_register_done();
 }
 module_exit(intel_rapl_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4672019..06d77fd 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -8,6 +8,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_PREPARE,
 	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
+	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -34,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/intel/rapl: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  8b5b773d6245138c461b3a3e242339fa50fa9c58
Gitweb:     http://git.kernel.org/tip/8b5b773d6245138c461b3a3e242339fa50fa9c58
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:15 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:34 +0200

perf/x86/intel/rapl: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.008808086@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/intel/rapl.c | 84 +++++++++++++-------------------------------
 include/linux/cpuhotplug.h   |  2 ++
 2 files changed, 27 insertions(+), 59 deletions(-)

diff --git a/arch/x86/events/intel/rapl.c b/arch/x86/events/intel/rapl.c
index e30eef4..6255ede 100644
--- a/arch/x86/events/intel/rapl.c
+++ b/arch/x86/events/intel/rapl.c
@@ -555,14 +555,14 @@ const struct attribute_group *rapl_attr_groups[] = {
 	NULL,
 };
 
-static void rapl_cpu_exit(int cpu)
+static int rapl_cpu_offline(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
 
 	/* Check if exiting cpu is used for collecting rapl events */
 	if (!cpumask_test_and_clear_cpu(cpu, &rapl_cpu_mask))
-		return;
+		return 0;
 
 	pmu->cpu = -1;
 	/* Find a new cpu to collect rapl events */
@@ -574,9 +574,10 @@ static void rapl_cpu_exit(int cpu)
 		pmu->cpu = target;
 		perf_pmu_migrate_context(pmu->pmu, cpu, target);
 	}
+	return 0;
 }
 
-static void rapl_cpu_init(int cpu)
+static int rapl_cpu_online(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 	int target;
@@ -587,13 +588,14 @@ static void rapl_cpu_init(int cpu)
 	 */
 	target = cpumask_any_and(&rapl_cpu_mask, topology_core_cpumask(cpu));
 	if (target < nr_cpu_ids)
-		return;
+		return 0;
 
 	cpumask_set_cpu(cpu, &rapl_cpu_mask);
 	pmu->cpu = cpu;
+	return 0;
 }
 
-static int rapl_cpu_prepare(int cpu)
+static int rapl_cpu_prepare(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
 
@@ -614,33 +616,6 @@ static int rapl_cpu_prepare(int cpu)
 	return 0;
 }
 
-static int rapl_cpu_notifier(struct notifier_block *self,
-			     unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		rapl_cpu_prepare(cpu);
-		break;
-
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		rapl_cpu_init(cpu);
-		break;
-
-	case CPU_DOWN_PREPARE:
-		rapl_cpu_exit(cpu);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block rapl_cpu_nb = {
-	.notifier_call	= rapl_cpu_notifier,
-	.priority       = CPU_PRI_PERF + 1,
-};
-
 static int rapl_check_hw_unit(bool apply_quirk)
 {
 	u64 msr_rapl_power_unit_bits;
@@ -691,24 +666,6 @@ static void __init rapl_advertise(void)
 	}
 }
 
-static int __init rapl_prepare_cpus(void)
-{
-	unsigned int cpu, pkg;
-	int ret;
-
-	for_each_online_cpu(cpu) {
-		pkg = topology_logical_package_id(cpu);
-		if (rapl_pmus->pmus[pkg])
-			continue;
-
-		ret = rapl_cpu_prepare(cpu);
-		if (ret)
-			return ret;
-		rapl_cpu_init(cpu);
-	}
-	return 0;
-}
-
 static void cleanup_rapl_pmus(void)
 {
 	int i;
@@ -835,35 +792,44 @@ static int __init rapl_pmu_init(void)
 	if (ret)
 		return ret;
 
-	cpu_notifier_register_begin();
+	/*
+	 * Install callbacks. Core will call them for each online cpu.
+	 */
 
-	ret = rapl_prepare_cpus();
+	ret = cpuhp_setup_state(CPUHP_PERF_X86_RAPL_PREP, "PERF_X86_RAPL_PREP",
+				rapl_cpu_prepare, NULL);
 	if (ret)
 		goto out;
 
+	ret = cpuhp_setup_state(CPUHP_AP_PERF_X86_RAPL_ONLINE,
+				"AP_PERF_X86_RAPL_ONLINE",
+				rapl_cpu_online, rapl_cpu_offline);
+	if (ret)
+		goto out1;
+
 	ret = perf_pmu_register(&rapl_pmus->pmu, "power", -1);
 	if (ret)
-		goto out;
+		goto out2;
 
-	__register_cpu_notifier(&rapl_cpu_nb);
-	cpu_notifier_register_done();
 	rapl_advertise();
 	return 0;
 
+out2:
+	cpuhp_remove_state(CPUHP_AP_PERF_X86_RAPL_ONLINE);
+out1:
+	cpuhp_remove_state(CPUHP_PERF_X86_RAPL_PREP);
 out:
 	pr_warn("Initialization failed (%d), disabled\n", ret);
 	cleanup_rapl_pmus();
-	cpu_notifier_register_done();
 	return ret;
 }
 module_init(rapl_pmu_init);
 
 static void __exit intel_rapl_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&rapl_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_RAPL_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_PERF_X86_RAPL_PREP);
 	perf_pmu_unregister(&rapl_pmus->pmu);
 	cleanup_rapl_pmus();
-	cpu_notifier_register_done();
 }
 module_exit(intel_rapl_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4672019..06d77fd 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -8,6 +8,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_PREPARE,
 	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
+	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -34,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 15/67] perf/x86/intel/cqm: Convert Intel CQM to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0014-perf-x86-intel-cqm-Convert-Intel-CQM-to-hotplug-stat.patch --]
[-- Type: text/plain, Size: 4710 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vikas Shivappa <vikas.shivappa@linux.intel.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/intel/cqm.c | 49 +++++++++++++++++----------------------------
 include/linux/cpuhotplug.h  |  2 ++
 2 files changed, 20 insertions(+), 31 deletions(-)

diff --git a/arch/x86/events/intel/cqm.c b/arch/x86/events/intel/cqm.c
index 7b5fd81..783c49d 100644
--- a/arch/x86/events/intel/cqm.c
+++ b/arch/x86/events/intel/cqm.c
@@ -1577,7 +1577,7 @@ static inline void cqm_pick_event_reader(int cpu)
 		cpumask_set_cpu(cpu, &cqm_cpumask);
 }
 
-static void intel_cqm_cpu_starting(unsigned int cpu)
+static int intel_cqm_cpu_starting(unsigned int cpu)
 {
 	struct intel_pqr_state *state = &per_cpu(pqr_state, cpu);
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
@@ -1588,39 +1588,26 @@ static void intel_cqm_cpu_starting(unsigned int cpu)
 
 	WARN_ON(c->x86_cache_max_rmid != cqm_max_rmid);
 	WARN_ON(c->x86_cache_occ_scale != cqm_l3_scale);
+
+	cqm_pick_event_reader(cpu);
+	return 0;
 }
 
-static void intel_cqm_cpu_exit(unsigned int cpu)
+static int intel_cqm_cpu_exit(unsigned int cpu)
 {
 	int target;
 
 	/* Is @cpu the current cqm reader for this package ? */
 	if (!cpumask_test_and_clear_cpu(cpu, &cqm_cpumask))
-		return;
+		return 0;
 
 	/* Find another online reader in this package */
 	target = cpumask_any_but(topology_core_cpumask(cpu), cpu);
 
 	if (target < nr_cpu_ids)
 		cpumask_set_cpu(target, &cqm_cpumask);
-}
-
-static int intel_cqm_cpu_notifier(struct notifier_block *nb,
-				  unsigned long action, void *hcpu)
-{
-	unsigned int cpu  = (unsigned long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
-		intel_cqm_cpu_exit(cpu);
-		break;
-	case CPU_STARTING:
-		intel_cqm_cpu_starting(cpu);
-		cqm_pick_event_reader(cpu);
-		break;
-	}
 
-	return NOTIFY_OK;
+	return 0;
 }
 
 static const struct x86_cpu_id intel_cqm_match[] = {
@@ -1682,7 +1669,7 @@ out:
 static int __init intel_cqm_init(void)
 {
 	char *str = NULL, scale[20];
-	int i, cpu, ret;
+	int cpu, ret;
 
 	if (x86_match_cpu(intel_cqm_match))
 		cqm_enabled = true;
@@ -1705,8 +1692,7 @@ static int __init intel_cqm_init(void)
 	 *
 	 * Also, check that the scales match on all cpus.
 	 */
-	cpu_notifier_register_begin();
-
+	get_online_cpus();
 	for_each_online_cpu(cpu) {
 		struct cpuinfo_x86 *c = &cpu_data(cpu);
 
@@ -1743,11 +1729,6 @@ static int __init intel_cqm_init(void)
 	if (ret)
 		goto out;
 
-	for_each_online_cpu(i) {
-		intel_cqm_cpu_starting(i);
-		cqm_pick_event_reader(i);
-	}
-
 	if (mbm_enabled)
 		ret = intel_mbm_init();
 	if (ret && !cqm_enabled)
@@ -1772,12 +1753,18 @@ static int __init intel_cqm_init(void)
 		pr_info("Intel MBM enabled\n");
 
 	/*
-	 * Register the hot cpu notifier once we are sure cqm
+	 * Setup the hot cpu notifier once we are sure cqm
 	 * is enabled to avoid notifier leak.
 	 */
-	__perf_cpu_notifier(intel_cqm_cpu_notifier);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CQM_STARTING,
+			  "AP_PERF_X86_CQM_STARTING",
+			  intel_cqm_cpu_starting, NULL);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CQM_ONLINE, "AP_PERF_X86_CQM_ONLINE",
+			  NULL, intel_cqm_cpu_exit);
+
 out:
-	cpu_notifier_register_done();
+	put_online_cpus();
+
 	if (ret) {
 		kfree(str);
 		cqm_cleanup();
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 06d77fd..f9399ee 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -25,6 +25,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
+	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -36,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
+	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/intel/cqm: Convert Intel CQM to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  f07048270423622a2428a9b90929c76e92777caa
Gitweb:     http://git.kernel.org/tip/f07048270423622a2428a9b90929c76e92777caa
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:16 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:35 +0200

perf/x86/intel/cqm: Convert Intel CQM to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vikas Shivappa <vikas.shivappa@linux.intel.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.096956222@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/intel/cqm.c | 49 +++++++++++++++++----------------------------
 include/linux/cpuhotplug.h  |  2 ++
 2 files changed, 20 insertions(+), 31 deletions(-)

diff --git a/arch/x86/events/intel/cqm.c b/arch/x86/events/intel/cqm.c
index 7b5fd81..783c49d 100644
--- a/arch/x86/events/intel/cqm.c
+++ b/arch/x86/events/intel/cqm.c
@@ -1577,7 +1577,7 @@ static inline void cqm_pick_event_reader(int cpu)
 		cpumask_set_cpu(cpu, &cqm_cpumask);
 }
 
-static void intel_cqm_cpu_starting(unsigned int cpu)
+static int intel_cqm_cpu_starting(unsigned int cpu)
 {
 	struct intel_pqr_state *state = &per_cpu(pqr_state, cpu);
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
@@ -1588,39 +1588,26 @@ static void intel_cqm_cpu_starting(unsigned int cpu)
 
 	WARN_ON(c->x86_cache_max_rmid != cqm_max_rmid);
 	WARN_ON(c->x86_cache_occ_scale != cqm_l3_scale);
+
+	cqm_pick_event_reader(cpu);
+	return 0;
 }
 
-static void intel_cqm_cpu_exit(unsigned int cpu)
+static int intel_cqm_cpu_exit(unsigned int cpu)
 {
 	int target;
 
 	/* Is @cpu the current cqm reader for this package ? */
 	if (!cpumask_test_and_clear_cpu(cpu, &cqm_cpumask))
-		return;
+		return 0;
 
 	/* Find another online reader in this package */
 	target = cpumask_any_but(topology_core_cpumask(cpu), cpu);
 
 	if (target < nr_cpu_ids)
 		cpumask_set_cpu(target, &cqm_cpumask);
-}
-
-static int intel_cqm_cpu_notifier(struct notifier_block *nb,
-				  unsigned long action, void *hcpu)
-{
-	unsigned int cpu  = (unsigned long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
-		intel_cqm_cpu_exit(cpu);
-		break;
-	case CPU_STARTING:
-		intel_cqm_cpu_starting(cpu);
-		cqm_pick_event_reader(cpu);
-		break;
-	}
 
-	return NOTIFY_OK;
+	return 0;
 }
 
 static const struct x86_cpu_id intel_cqm_match[] = {
@@ -1682,7 +1669,7 @@ out:
 static int __init intel_cqm_init(void)
 {
 	char *str = NULL, scale[20];
-	int i, cpu, ret;
+	int cpu, ret;
 
 	if (x86_match_cpu(intel_cqm_match))
 		cqm_enabled = true;
@@ -1705,8 +1692,7 @@ static int __init intel_cqm_init(void)
 	 *
 	 * Also, check that the scales match on all cpus.
 	 */
-	cpu_notifier_register_begin();
-
+	get_online_cpus();
 	for_each_online_cpu(cpu) {
 		struct cpuinfo_x86 *c = &cpu_data(cpu);
 
@@ -1743,11 +1729,6 @@ static int __init intel_cqm_init(void)
 	if (ret)
 		goto out;
 
-	for_each_online_cpu(i) {
-		intel_cqm_cpu_starting(i);
-		cqm_pick_event_reader(i);
-	}
-
 	if (mbm_enabled)
 		ret = intel_mbm_init();
 	if (ret && !cqm_enabled)
@@ -1772,12 +1753,18 @@ static int __init intel_cqm_init(void)
 		pr_info("Intel MBM enabled\n");
 
 	/*
-	 * Register the hot cpu notifier once we are sure cqm
+	 * Setup the hot cpu notifier once we are sure cqm
 	 * is enabled to avoid notifier leak.
 	 */
-	__perf_cpu_notifier(intel_cqm_cpu_notifier);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CQM_STARTING,
+			  "AP_PERF_X86_CQM_STARTING",
+			  intel_cqm_cpu_starting, NULL);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CQM_ONLINE, "AP_PERF_X86_CQM_ONLINE",
+			  NULL, intel_cqm_cpu_exit);
+
 out:
-	cpu_notifier_register_done();
+	put_online_cpus();
+
 	if (ret) {
 		kfree(str);
 		cqm_cleanup();
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 06d77fd..f9399ee 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -25,6 +25,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
+	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -36,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
+	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 16/67] perf/x86/intel/cstate: Convert Intel CSTATE to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0015-perf-x86-intel-cstate-Convert-Intel-CSTATE-to-hotplu.patch --]
[-- Type: text/plain, Size: 4619 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/events/intel/cstate.c | 51 +++++++++++++-----------------------------
 include/linux/cpuhotplug.h     |  2 ++
 2 files changed, 17 insertions(+), 36 deletions(-)

diff --git a/arch/x86/events/intel/cstate.c b/arch/x86/events/intel/cstate.c
index 9ba4e41..d6d7be0 100644
--- a/arch/x86/events/intel/cstate.c
+++ b/arch/x86/events/intel/cstate.c
@@ -365,7 +365,7 @@ static int cstate_pmu_event_add(struct perf_event *event, int mode)
  * Check if exiting cpu is the designated reader. If so migrate the
  * events when there is a valid target available
  */
-static void cstate_cpu_exit(int cpu)
+static int cstate_cpu_exit(unsigned int cpu)
 {
 	unsigned int target;
 
@@ -390,9 +390,10 @@ static void cstate_cpu_exit(int cpu)
 			perf_pmu_migrate_context(&cstate_pkg_pmu, cpu, target);
 		}
 	}
+	return 0;
 }
 
-static void cstate_cpu_init(int cpu)
+static int cstate_cpu_init(unsigned int cpu)
 {
 	unsigned int target;
 
@@ -414,31 +415,10 @@ static void cstate_cpu_init(int cpu)
 				 topology_core_cpumask(cpu));
 	if (has_cstate_pkg && target >= nr_cpu_ids)
 		cpumask_set_cpu(cpu, &cstate_pkg_cpu_mask);
-}
 
-static int cstate_cpu_notifier(struct notifier_block *self,
-			       unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		cstate_cpu_init(cpu);
-		break;
-	case CPU_DOWN_PREPARE:
-		cstate_cpu_exit(cpu);
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block cstate_cpu_nb = {
-	.notifier_call	= cstate_cpu_notifier,
-	.priority       = CPU_PRI_PERF + 1,
-};
-
 static struct pmu cstate_core_pmu = {
 	.attr_groups	= core_attr_groups,
 	.name		= "cstate_core",
@@ -599,18 +579,20 @@ static inline void cstate_cleanup(void)
 
 static int __init cstate_init(void)
 {
-	int cpu, err;
+	int err;
 
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu)
-		cstate_cpu_init(cpu);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CSTATE_STARTING,
+			  "AP_PERF_X86_CSTATE_STARTING", cstate_cpu_init,
+			  NULL);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CSTATE_ONLINE,
+			  "AP_PERF_X86_CSTATE_ONLINE", NULL, cstate_cpu_exit);
 
 	if (has_cstate_core) {
 		err = perf_pmu_register(&cstate_core_pmu, cstate_core_pmu.name, -1);
 		if (err) {
 			has_cstate_core = false;
 			pr_info("Failed to register cstate core pmu\n");
-			goto out;
+			return err;
 		}
 	}
 
@@ -620,12 +602,10 @@ static int __init cstate_init(void)
 			has_cstate_pkg = false;
 			pr_info("Failed to register cstate pkg pmu\n");
 			cstate_cleanup();
-			goto out;
+			return err;
 		}
 	}
-	__register_cpu_notifier(&cstate_cpu_nb);
-out:
-	cpu_notifier_register_done();
+
 	return err;
 }
 
@@ -651,9 +631,8 @@ module_init(cstate_pmu_init);
 
 static void __exit cstate_pmu_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&cstate_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_CSTATE_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_CSTATE_STARTING);
 	cstate_cleanup();
-	cpu_notifier_register_done();
 }
 module_exit(cstate_pmu_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f9399ee..68f4495 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -26,6 +26,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_PERF_X86_CQM_STARTING,
+	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -38,6 +39,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
+	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/intel/cstate: Convert Intel CSTATE to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  77c34ef1c3194bfac65883af75baf7dec9fa0d77
Gitweb:     http://git.kernel.org/tip/77c34ef1c3194bfac65883af75baf7dec9fa0d77
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:18 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:35 +0200

perf/x86/intel/cstate: Convert Intel CSTATE to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: kbuild test robot <fengguang.wu@intel.com>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.184061086@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/intel/cstate.c | 51 +++++++++++++-----------------------------
 include/linux/cpuhotplug.h     |  2 ++
 2 files changed, 17 insertions(+), 36 deletions(-)

diff --git a/arch/x86/events/intel/cstate.c b/arch/x86/events/intel/cstate.c
index 9ba4e41..d6d7be0 100644
--- a/arch/x86/events/intel/cstate.c
+++ b/arch/x86/events/intel/cstate.c
@@ -365,7 +365,7 @@ static int cstate_pmu_event_add(struct perf_event *event, int mode)
  * Check if exiting cpu is the designated reader. If so migrate the
  * events when there is a valid target available
  */
-static void cstate_cpu_exit(int cpu)
+static int cstate_cpu_exit(unsigned int cpu)
 {
 	unsigned int target;
 
@@ -390,9 +390,10 @@ static void cstate_cpu_exit(int cpu)
 			perf_pmu_migrate_context(&cstate_pkg_pmu, cpu, target);
 		}
 	}
+	return 0;
 }
 
-static void cstate_cpu_init(int cpu)
+static int cstate_cpu_init(unsigned int cpu)
 {
 	unsigned int target;
 
@@ -414,31 +415,10 @@ static void cstate_cpu_init(int cpu)
 				 topology_core_cpumask(cpu));
 	if (has_cstate_pkg && target >= nr_cpu_ids)
 		cpumask_set_cpu(cpu, &cstate_pkg_cpu_mask);
-}
 
-static int cstate_cpu_notifier(struct notifier_block *self,
-			       unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		cstate_cpu_init(cpu);
-		break;
-	case CPU_DOWN_PREPARE:
-		cstate_cpu_exit(cpu);
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block cstate_cpu_nb = {
-	.notifier_call	= cstate_cpu_notifier,
-	.priority       = CPU_PRI_PERF + 1,
-};
-
 static struct pmu cstate_core_pmu = {
 	.attr_groups	= core_attr_groups,
 	.name		= "cstate_core",
@@ -599,18 +579,20 @@ static inline void cstate_cleanup(void)
 
 static int __init cstate_init(void)
 {
-	int cpu, err;
+	int err;
 
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu)
-		cstate_cpu_init(cpu);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CSTATE_STARTING,
+			  "AP_PERF_X86_CSTATE_STARTING", cstate_cpu_init,
+			  NULL);
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_CSTATE_ONLINE,
+			  "AP_PERF_X86_CSTATE_ONLINE", NULL, cstate_cpu_exit);
 
 	if (has_cstate_core) {
 		err = perf_pmu_register(&cstate_core_pmu, cstate_core_pmu.name, -1);
 		if (err) {
 			has_cstate_core = false;
 			pr_info("Failed to register cstate core pmu\n");
-			goto out;
+			return err;
 		}
 	}
 
@@ -620,12 +602,10 @@ static int __init cstate_init(void)
 			has_cstate_pkg = false;
 			pr_info("Failed to register cstate pkg pmu\n");
 			cstate_cleanup();
-			goto out;
+			return err;
 		}
 	}
-	__register_cpu_notifier(&cstate_cpu_nb);
-out:
-	cpu_notifier_register_done();
+
 	return err;
 }
 
@@ -651,9 +631,8 @@ module_init(cstate_pmu_init);
 
 static void __exit cstate_pmu_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&cstate_cpu_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_CSTATE_ONLINE);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_CSTATE_STARTING);
 	cstate_cleanup();
-	cpu_notifier_register_done();
 }
 module_exit(cstate_pmu_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f9399ee..68f4495 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -26,6 +26,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_STARTING,
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_PERF_X86_CQM_STARTING,
+	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
@@ -38,6 +39,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
+	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 17/67] blackfin/perf: Convert hotplug notifier to state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0016-blackfin-perf-Convert-hotplug-notifier-to-state-mach.patch --]
[-- Type: text/plain, Size: 2375 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callback via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Miao <realmz6@gmail.com>
Cc: adi-buildroot-devel@lists.sourceforge.net
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/blackfin/kernel/perf_event.c | 26 +++++---------------------
 include/linux/cpuhotplug.h        |  1 +
 2 files changed, 6 insertions(+), 21 deletions(-)

diff --git a/arch/blackfin/kernel/perf_event.c b/arch/blackfin/kernel/perf_event.c
index 170d786..6355e97 100644
--- a/arch/blackfin/kernel/perf_event.c
+++ b/arch/blackfin/kernel/perf_event.c
@@ -453,29 +453,13 @@ static struct pmu pmu = {
 	.read        = bfin_pmu_read,
 };
 
-static void bfin_pmu_setup(int cpu)
+static int bfin_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
+	bfin_write_PFCTL(0);
 	memset(cpuhw, 0, sizeof(struct cpu_hw_events));
-}
-
-static int
-bfin_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		bfin_write_PFCTL(0);
-		bfin_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 static int __init bfin_pmu_init(void)
@@ -491,8 +475,8 @@ static int __init bfin_pmu_init(void)
 
 	ret = perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
 	if (!ret)
-		perf_cpu_notifier(bfin_pmu_notifier);
-
+		cpuhp_setup_state(CPUHP_PERF_BFIN, "PERF_BFIN",
+				  bfin_pmu_prepare_cpu, NULL);
 	return ret;
 }
 early_initcall(bfin_pmu_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 68f4495..d54973e 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -9,6 +9,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_PERF_X86_RAPL_PREP,
+	CPUHP_PERF_BFIN,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
-- 
2.8.1

[tip:smp/hotplug] blackfin/perf: Convert hotplug notifier to state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  a409f5ee2937cb15b4af485fa98b90946d9bea3a
Gitweb:     http://git.kernel.org/tip/a409f5ee2937cb15b4af485fa98b90946d9bea3a
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:19 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:36 +0200

blackfin/perf: Convert hotplug notifier to state machine

Install the callback via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Miao <realmz6@gmail.com>
Cc: adi-buildroot-devel@lists.sourceforge.net
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.265797537@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/blackfin/kernel/perf_event.c | 26 +++++---------------------
 include/linux/cpuhotplug.h        |  1 +
 2 files changed, 6 insertions(+), 21 deletions(-)

diff --git a/arch/blackfin/kernel/perf_event.c b/arch/blackfin/kernel/perf_event.c
index 170d786..6355e97 100644
--- a/arch/blackfin/kernel/perf_event.c
+++ b/arch/blackfin/kernel/perf_event.c
@@ -453,29 +453,13 @@ static struct pmu pmu = {
 	.read        = bfin_pmu_read,
 };
 
-static void bfin_pmu_setup(int cpu)
+static int bfin_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
+	bfin_write_PFCTL(0);
 	memset(cpuhw, 0, sizeof(struct cpu_hw_events));
-}
-
-static int
-bfin_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		bfin_write_PFCTL(0);
-		bfin_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 static int __init bfin_pmu_init(void)
@@ -491,8 +475,8 @@ static int __init bfin_pmu_init(void)
 
 	ret = perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
 	if (!ret)
-		perf_cpu_notifier(bfin_pmu_notifier);
-
+		cpuhp_setup_state(CPUHP_PERF_BFIN, "PERF_BFIN",
+				  bfin_pmu_prepare_cpu, NULL);
 	return ret;
 }
 early_initcall(bfin_pmu_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 68f4495..d54973e 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -9,6 +9,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_UNCORE_PREP,
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_PERF_X86_RAPL_PREP,
+	CPUHP_PERF_BFIN,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,

[patch V2 18/67] powerpc/perf: Convert book3s notifier to state machine callbacks

[Anna-Maria Gleixner]

[-- Attachment #1: 0017-powerpc-perf-Convert-book3s-notifier-to-state-machin.patch --]
[-- Type: text/plain, Size: 2545 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/powerpc/perf/core-book3s.c | 30 +++++++-----------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 8 insertions(+), 23 deletions(-)

diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 97a1d40..ffd61d5 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -2158,31 +2158,15 @@ static void perf_event_interrupt(struct pt_regs *regs)
 		irq_exit();
 }
 
-static void power_pmu_setup(int cpu)
+int power_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
-	if (!ppmu)
-		return;
-	memset(cpuhw, 0, sizeof(*cpuhw));
-	cpuhw->mmcr[0] = MMCR0_FC;
-}
-
-static int
-power_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		power_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
+	if (ppmu) {
+		memset(cpuhw, 0, sizeof(*cpuhw));
+		cpuhw->mmcr[0] = MMCR0_FC;
 	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 int register_power_pmu(struct power_pmu *pmu)
@@ -2205,7 +2189,7 @@ int register_power_pmu(struct power_pmu *pmu)
 #endif /* CONFIG_PPC64 */
 
 	perf_pmu_register(&power_pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(power_pmu_notifier);
-
+	cpuhp_setup_state(CPUHP_PERF_POWER, "PERF_POWER",
+			  power_pmu_prepare_cpu, NULL);
 	return 0;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d54973e..b265e99 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -10,6 +10,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_PERF_BFIN,
+	CPUHP_PERF_POWER,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
-- 
2.8.1

[tip:smp/hotplug] powerpc/perf: Convert book3s notifier to state machine callbacks

[tip-bot for Thomas Gleixner]

Commit-ID:  57ecde42cc7493fb74029918d0a8f97aec854876
Gitweb:     http://git.kernel.org/tip/57ecde42cc7493fb74029918d0a8f97aec854876
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:20 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:37 +0200

powerpc/perf: Convert book3s notifier to state machine callbacks

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.345786236@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/powerpc/perf/core-book3s.c | 30 +++++++-----------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 8 insertions(+), 23 deletions(-)

diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 97a1d40..ffd61d5 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -2158,31 +2158,15 @@ static void perf_event_interrupt(struct pt_regs *regs)
 		irq_exit();
 }
 
-static void power_pmu_setup(int cpu)
+int power_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
-	if (!ppmu)
-		return;
-	memset(cpuhw, 0, sizeof(*cpuhw));
-	cpuhw->mmcr[0] = MMCR0_FC;
-}
-
-static int
-power_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		power_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
+	if (ppmu) {
+		memset(cpuhw, 0, sizeof(*cpuhw));
+		cpuhw->mmcr[0] = MMCR0_FC;
 	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 int register_power_pmu(struct power_pmu *pmu)
@@ -2205,7 +2189,7 @@ int register_power_pmu(struct power_pmu *pmu)
 #endif /* CONFIG_PPC64 */
 
 	perf_pmu_register(&power_pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(power_pmu_notifier);
-
+	cpuhp_setup_state(CPUHP_PERF_POWER, "PERF_POWER",
+			  power_pmu_prepare_cpu, NULL);
 	return 0;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d54973e..b265e99 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -10,6 +10,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_AMD_UNCORE_PREP,
 	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_PERF_BFIN,
+	CPUHP_PERF_POWER,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,

[patch V2 19/67] s390/perf: Convert the hotplug notifier to state machine callbacks (Counter)

[Anna-Maria Gleixner]

[-- Attachment #1: 0018-s390-perf-Convert-the-hotplug-notifier-to-state-mach.patch --]
[-- Type: text/plain, Size: 3239 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Adam Buchbinder <adam.buchbinder@gmail.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-s390@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/s390/kernel/perf_cpum_cf.c | 46 +++++++++++++++++------------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 20 insertions(+), 27 deletions(-)

diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c
index 7ec63b1..037c2a2 100644
--- a/arch/s390/kernel/perf_cpum_cf.c
+++ b/arch/s390/kernel/perf_cpum_cf.c
@@ -664,30 +664,22 @@ static struct pmu cpumf_pmu = {
 	.cancel_txn   = cpumf_pmu_cancel_txn,
 };
 
-static int cpumf_pmu_notifier(struct notifier_block *self, unsigned long action,
-			      void *hcpu)
+static int cpumf_pmf_setup(unsigned int cpu, int flags)
 {
-	int flags;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		flags = PMC_INIT;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	case CPU_DOWN_PREPARE:
-		flags = PMC_RELEASE;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	default:
-		break;
-	}
+	local_irq_disable();
+	setup_pmc_cpu(&flags);
+	local_irq_enable();
+	return 0;
+}
+
+static int s390_pmu_online_cpu(unsigned int cpu)
+{
+	return cpumf_pmf_setup(cpu, PMC_INIT);
+}
 
-	return NOTIFY_OK;
+static int s390_pmu_offline_cpu(unsigned int cpu)
+{
+	return cpumf_pmf_setup(cpu, PMC_RELEASE);
 }
 
 static int __init cpumf_pmu_init(void)
@@ -707,7 +699,7 @@ static int __init cpumf_pmu_init(void)
 	if (rc) {
 		pr_err("Registering for CPU-measurement alerts "
 		       "failed with rc=%i\n", rc);
-		goto out;
+		return rc;
 	}
 
 	cpumf_pmu.attr_groups = cpumf_cf_event_group();
@@ -716,10 +708,10 @@ static int __init cpumf_pmu_init(void)
 		pr_err("Registering the cpum_cf PMU failed with rc=%i\n", rc);
 		unregister_external_irq(EXT_IRQ_MEASURE_ALERT,
 					cpumf_measurement_alert);
-		goto out;
+		return rc;
 	}
-	perf_cpu_notifier(cpumf_pmu_notifier);
-out:
-	return rc;
+	return cpuhp_setup_state(CPUHP_AP_PERF_S390_CF_ONLINE,
+				 "AP_PERF_S390_CF_ONLINE",
+				 s390_pmu_online_cpu, s390_pmu_offline_cpu);
 }
 early_initcall(cpumf_pmu_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index b265e99..d4e09cb 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -42,6 +42,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
+	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] s390/perf: Convert the hotplug notifier to state machine callbacks (Counter)

[tip-bot for Thomas Gleixner]

Commit-ID:  4f0f8217e69a0f6ff727913aa49cb64632ff7b6a
Gitweb:     http://git.kernel.org/tip/4f0f8217e69a0f6ff727913aa49cb64632ff7b6a
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:21 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:37 +0200

s390/perf: Convert the hotplug notifier to state machine callbacks (Counter)

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Adam Buchbinder <adam.buchbinder@gmail.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: linux-s390@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.436370635@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/s390/kernel/perf_cpum_cf.c | 46 +++++++++++++++++------------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 20 insertions(+), 27 deletions(-)

diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c
index 7ec63b1..037c2a2 100644
--- a/arch/s390/kernel/perf_cpum_cf.c
+++ b/arch/s390/kernel/perf_cpum_cf.c
@@ -664,30 +664,22 @@ static struct pmu cpumf_pmu = {
 	.cancel_txn   = cpumf_pmu_cancel_txn,
 };
 
-static int cpumf_pmu_notifier(struct notifier_block *self, unsigned long action,
-			      void *hcpu)
+static int cpumf_pmf_setup(unsigned int cpu, int flags)
 {
-	int flags;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		flags = PMC_INIT;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	case CPU_DOWN_PREPARE:
-		flags = PMC_RELEASE;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	default:
-		break;
-	}
+	local_irq_disable();
+	setup_pmc_cpu(&flags);
+	local_irq_enable();
+	return 0;
+}
+
+static int s390_pmu_online_cpu(unsigned int cpu)
+{
+	return cpumf_pmf_setup(cpu, PMC_INIT);
+}
 
-	return NOTIFY_OK;
+static int s390_pmu_offline_cpu(unsigned int cpu)
+{
+	return cpumf_pmf_setup(cpu, PMC_RELEASE);
 }
 
 static int __init cpumf_pmu_init(void)
@@ -707,7 +699,7 @@ static int __init cpumf_pmu_init(void)
 	if (rc) {
 		pr_err("Registering for CPU-measurement alerts "
 		       "failed with rc=%i\n", rc);
-		goto out;
+		return rc;
 	}
 
 	cpumf_pmu.attr_groups = cpumf_cf_event_group();
@@ -716,10 +708,10 @@ static int __init cpumf_pmu_init(void)
 		pr_err("Registering the cpum_cf PMU failed with rc=%i\n", rc);
 		unregister_external_irq(EXT_IRQ_MEASURE_ALERT,
 					cpumf_measurement_alert);
-		goto out;
+		return rc;
 	}
-	perf_cpu_notifier(cpumf_pmu_notifier);
-out:
-	return rc;
+	return cpuhp_setup_state(CPUHP_AP_PERF_S390_CF_ONLINE,
+				 "AP_PERF_S390_CF_ONLINE",
+				 s390_pmu_online_cpu, s390_pmu_offline_cpu);
 }
 early_initcall(cpumf_pmu_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index b265e99..d4e09cb 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -42,6 +42,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
+	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 20/67] s390/perf: Convert the hotplug notifier to state machine callbacks (Sampling)

[Anna-Maria Gleixner]

[-- Attachment #1: 0019-s390-perf-Convert-the-hotplug-notifier-to-state-mach.patch --]
[-- Type: text/plain, Size: 3015 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke the
callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-s390@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/s390/kernel/perf_cpum_sf.c | 45 +++++++++++++++++------------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 20 insertions(+), 26 deletions(-)

diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c
index a8e8321..f4a4c11 100644
--- a/arch/s390/kernel/perf_cpum_sf.c
+++ b/arch/s390/kernel/perf_cpum_sf.c
@@ -1506,37 +1506,28 @@ static void cpumf_measurement_alert(struct ext_code ext_code,
 		sf_disable();
 	}
 }
-
-static int cpumf_pmu_notifier(struct notifier_block *self,
-			      unsigned long action, void *hcpu)
+static int cpusf_pmu_setup(unsigned int cpu, int flags)
 {
-	int flags;
-
 	/* Ignore the notification if no events are scheduled on the PMU.
 	 * This might be racy...
 	 */
 	if (!atomic_read(&num_events))
-		return NOTIFY_OK;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		flags = PMC_INIT;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	case CPU_DOWN_PREPARE:
-		flags = PMC_RELEASE;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	default:
-		break;
-	}
+		return 0;
 
-	return NOTIFY_OK;
+	local_irq_disable();
+	setup_pmc_cpu(&flags);
+	local_irq_enable();
+	return 0;
+}
+
+static int s390_pmu_sf_online_cpu(unsigned int cpu)
+{
+	return cpusf_pmu_setup(cpu, PMC_INIT);
+}
+
+static int s390_pmu_sf_offline_cpu(unsigned int cpu)
+{
+	return cpusf_pmu_setup(cpu, PMC_RELEASE);
 }
 
 static int param_get_sfb_size(char *buffer, const struct kernel_param *kp)
@@ -1636,7 +1627,9 @@ static int __init init_cpum_sampling_pmu(void)
 					cpumf_measurement_alert);
 		goto out;
 	}
-	perf_cpu_notifier(cpumf_pmu_notifier);
+
+	cpuhp_setup_state(CPUHP_AP_PERF_S390_SF_ONLINE, "AP_PERF_S390_SF_ONLINE",
+			  s390_pmu_sf_online_cpu, s390_pmu_sf_offline_cpu);
 out:
 	return err;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d4e09cb..fea6666 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -43,6 +43,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_PERF_S390_CF_ONLINE,
+	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
-- 
2.8.1

[tip:smp/hotplug] s390/perf: Convert the hotplug notifier to state machine callbacks (Sampling)

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  e3d617fe6ac7294974fc513dc5e4d8ada8080fd1
Gitweb:     http://git.kernel.org/tip/e3d617fe6ac7294974fc513dc5e4d8ada8080fd1
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:22 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:38 +0200

s390/perf: Convert the hotplug notifier to state machine callbacks (Sampling)

Install the callbacks via the state machine and let the core invoke the
callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-s390@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.518084858@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/s390/kernel/perf_cpum_sf.c | 45 +++++++++++++++++------------------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 20 insertions(+), 26 deletions(-)

diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c
index a8e8321..f4a4c11 100644
--- a/arch/s390/kernel/perf_cpum_sf.c
+++ b/arch/s390/kernel/perf_cpum_sf.c
@@ -1506,37 +1506,28 @@ static void cpumf_measurement_alert(struct ext_code ext_code,
 		sf_disable();
 	}
 }
-
-static int cpumf_pmu_notifier(struct notifier_block *self,
-			      unsigned long action, void *hcpu)
+static int cpusf_pmu_setup(unsigned int cpu, int flags)
 {
-	int flags;
-
 	/* Ignore the notification if no events are scheduled on the PMU.
 	 * This might be racy...
 	 */
 	if (!atomic_read(&num_events))
-		return NOTIFY_OK;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		flags = PMC_INIT;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	case CPU_DOWN_PREPARE:
-		flags = PMC_RELEASE;
-		local_irq_disable();
-		setup_pmc_cpu(&flags);
-		local_irq_enable();
-		break;
-	default:
-		break;
-	}
+		return 0;
 
-	return NOTIFY_OK;
+	local_irq_disable();
+	setup_pmc_cpu(&flags);
+	local_irq_enable();
+	return 0;
+}
+
+static int s390_pmu_sf_online_cpu(unsigned int cpu)
+{
+	return cpusf_pmu_setup(cpu, PMC_INIT);
+}
+
+static int s390_pmu_sf_offline_cpu(unsigned int cpu)
+{
+	return cpusf_pmu_setup(cpu, PMC_RELEASE);
 }
 
 static int param_get_sfb_size(char *buffer, const struct kernel_param *kp)
@@ -1636,7 +1627,9 @@ static int __init init_cpum_sampling_pmu(void)
 					cpumf_measurement_alert);
 		goto out;
 	}
-	perf_cpu_notifier(cpumf_pmu_notifier);
+
+	cpuhp_setup_state(CPUHP_AP_PERF_S390_SF_ONLINE, "AP_PERF_S390_SF_ONLINE",
+			  s390_pmu_sf_online_cpu, s390_pmu_sf_offline_cpu);
 out:
 	return err;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d4e09cb..fea6666 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -43,6 +43,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_PERF_S390_CF_ONLINE,
+	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 21/67] sh/perf: Convert the hotplug notifiers to state machine callbacks

[Anna-Maria Gleixner]

[-- Attachment #1: 0020-sh-perf-Convert-the-hotplug-notifiers-to-state-machi.patch --]
[-- Type: text/plain, Size: 2242 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: linux-sh@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/sh/kernel/perf_event.c | 23 ++++-------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 5 insertions(+), 19 deletions(-)

diff --git a/arch/sh/kernel/perf_event.c b/arch/sh/kernel/perf_event.c
index 4dca183..ba3269a 100644
--- a/arch/sh/kernel/perf_event.c
+++ b/arch/sh/kernel/perf_event.c
@@ -352,28 +352,12 @@ static struct pmu pmu = {
 	.read		= sh_pmu_read,
 };
 
-static void sh_pmu_setup(int cpu)
+static int sh_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
 	memset(cpuhw, 0, sizeof(struct cpu_hw_events));
-}
-
-static int
-sh_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		sh_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 int register_sh_pmu(struct sh_pmu *_pmu)
@@ -394,6 +378,7 @@ int register_sh_pmu(struct sh_pmu *_pmu)
 	WARN_ON(_pmu->num_events > MAX_HWEVENTS);
 
 	perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(sh_pmu_notifier);
+	cpuhp_setup_state(CPUHP_PERF_SUPERH, "PERF_SUPERH", sh_pmu_prepare_cpu,
+			  NULL);
 	return 0;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index fea6666..16ab417 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -11,6 +11,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
+	CPUHP_PERF_SUPERH,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
-- 
2.8.1

[tip:smp/hotplug] sh/perf: Convert the hotplug notifiers to state machine callbacks

[tip-bot for Thomas Gleixner]

Commit-ID:  e3cfce17d309a3c914e36ac4b4b26c30045d4274
Gitweb:     http://git.kernel.org/tip/e3cfce17d309a3c914e36ac4b4b26c30045d4274
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:23 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:39 +0200

sh/perf: Convert the hotplug notifiers to state machine callbacks

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: linux-sh@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.597790464@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/sh/kernel/perf_event.c | 23 ++++-------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 5 insertions(+), 19 deletions(-)

diff --git a/arch/sh/kernel/perf_event.c b/arch/sh/kernel/perf_event.c
index 4dca183..ba3269a 100644
--- a/arch/sh/kernel/perf_event.c
+++ b/arch/sh/kernel/perf_event.c
@@ -352,28 +352,12 @@ static struct pmu pmu = {
 	.read		= sh_pmu_read,
 };
 
-static void sh_pmu_setup(int cpu)
+static int sh_pmu_prepare_cpu(unsigned int cpu)
 {
 	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
 
 	memset(cpuhw, 0, sizeof(struct cpu_hw_events));
-}
-
-static int
-sh_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		sh_pmu_setup(cpu);
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 int register_sh_pmu(struct sh_pmu *_pmu)
@@ -394,6 +378,7 @@ int register_sh_pmu(struct sh_pmu *_pmu)
 	WARN_ON(_pmu->num_events > MAX_HWEVENTS);
 
 	perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
-	perf_cpu_notifier(sh_pmu_notifier);
+	cpuhp_setup_state(CPUHP_PERF_SUPERH, "PERF_SUPERH", sh_pmu_prepare_cpu,
+			  NULL);
 	return 0;
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index fea6666..16ab417 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -11,6 +11,7 @@ enum cpuhp_state {
 	CPUHP_PERF_X86_RAPL_PREP,
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
+	CPUHP_PERF_SUPERH,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,

[patch V2 22/67] bus/arm-cci: Convert to hotplug statemachine

[Anna-Maria Gleixner]

[-- Attachment #1: 0021-bus-arm-cci-Convert-to-hotplug-statemachine.patch --]
[-- Type: text/plain, Size: 3982 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andrzej Hajda <a.hajda@samsung.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Olof Johansson <olof@lixom.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Punit Agrawal <punit.agrawal@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Suzuki K. Poulose <suzuki.poulose@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/bus/arm-cci.c      |   53 +++++++++++++++++++--------------------------
 include/linux/cpuhotplug.h |    1 
 2 files changed, 24 insertions(+), 30 deletions(-)

--- a/drivers/bus/arm-cci.c
+++ b/drivers/bus/arm-cci.c
@@ -144,12 +144,15 @@ struct cci_pmu {
 	int num_cntrs;
 	atomic_t active_events;
 	struct mutex reserve_mutex;
-	struct notifier_block cpu_nb;
+	struct list_head entry;
 	cpumask_t cpus;
 };
 
 #define to_cci_pmu(c)	(container_of(c, struct cci_pmu, pmu))
 
+static DEFINE_MUTEX(cci_pmu_mutex);
+static LIST_HEAD(cci_pmu_list);
+
 enum cci_models {
 #ifdef CONFIG_ARM_CCI400_PMU
 	CCI400_R0,
@@ -1503,31 +1506,26 @@ static int cci_pmu_init(struct cci_pmu *
 	return perf_pmu_register(&cci_pmu->pmu, name, -1);
 }
 
-static int cci_pmu_cpu_notifier(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+static int cci_pmu_offline_cpu(unsigned int cpu)
 {
-	struct cci_pmu *cci_pmu = container_of(self,
-					struct cci_pmu, cpu_nb);
-	unsigned int cpu = (long)hcpu;
+	struct cci_pmu *cci_pmu;
 	unsigned int target;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
+	mutex_lock(&cci_pmu_mutex);
+	list_for_each_entry(cci_pmu, &cci_pmu_list, entry) {
 		if (!cpumask_test_and_clear_cpu(cpu, &cci_pmu->cpus))
-			break;
+			continue;
 		target = cpumask_any_but(cpu_online_mask, cpu);
-		if (target >= nr_cpu_ids) // UP, last CPU
-			break;
+		if (target >= nr_cpu_ids)
+			continue;
 		/*
 		 * TODO: migrate context once core races on event->ctx have
 		 * been fixed.
 		 */
 		cpumask_set_cpu(target, &cci_pmu->cpus);
-	default:
-		break;
 	}
-
-	return NOTIFY_OK;
+	mutex_unlock(&cci_pmu_mutex);
+	return 0;
 }
 
 static struct cci_pmu_model cci_pmu_models[] = {
@@ -1766,24 +1764,13 @@ static int cci_pmu_probe(struct platform
 	atomic_set(&cci_pmu->active_events, 0);
 	cpumask_set_cpu(smp_processor_id(), &cci_pmu->cpus);
 
-	cci_pmu->cpu_nb = (struct notifier_block) {
-		.notifier_call	= cci_pmu_cpu_notifier,
-		/*
-		 * to migrate uncore events, our notifier should be executed
-		 * before perf core's notifier.
-		 */
-		.priority	= CPU_PRI_PERF + 1,
-	};
-
-	ret = register_cpu_notifier(&cci_pmu->cpu_nb);
+	ret = cci_pmu_init(cci_pmu, pdev);
 	if (ret)
 		return ret;
 
-	ret = cci_pmu_init(cci_pmu, pdev);
-	if (ret) {
-		unregister_cpu_notifier(&cci_pmu->cpu_nb);
-		return ret;
-	}
+	mutex_lock(&cci_pmu_mutex);
+	list_add(&cci_pmu->entry, &cci_pmu_list);
+	mutex_unlock(&cci_pmu_mutex);
 
 	pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
 	return 0;
@@ -1817,6 +1804,12 @@ static int __init cci_platform_init(void
 {
 	int ret;
 
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+					"AP_PERF_ARM_CCI_ONLINE", NULL,
+					cci_pmu_offline_cpu);
+	if (ret)
+		return ret;
+
 	ret = platform_driver_register(&cci_pmu_driver);
 	if (ret)
 		return ret;
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -45,6 +45,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_PERF_S390_SF_ONLINE,
+	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

Re: [patch V2 22/67] bus/arm-cci: Convert to hotplug statemachine

[Punit Agrawal]

Anna-Maria Gleixner <anna-maria@linutronix.de> writes:

> From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
>
> Install the callbacks via the state machine and let the core invoke
> the callbacks on the already online CPUs.
>
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Andrzej Hajda <a.hajda@samsung.com>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: Olof Johansson <olof@lixom.net>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Punit Agrawal <punit.agrawal@arm.com>
> Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
> Cc: Suzuki K. Poulose <suzuki.poulose@arm.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Will Deacon <will.deacon@arm.com>
> Cc: Pawel Moll <pawel.moll@arm.com>
> Cc: linux-kernel@vger.kernel.org
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>

We don't expect to have multiple CCIs in the system. I suspect this got
changed from v1 along with ccn. Even then the patch still does the right
thing.

Acked-by: Punit Agrawal <punit.agrawal@arm.com>

Thanks.


[...]

[tip:smp/hotplug] bus/arm-cci: Convert to hotplug statemachine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  28c94843bb873196f704e9a26775b117a32a31ad
Gitweb:     http://git.kernel.org/tip/28c94843bb873196f704e9a26775b117a32a31ad
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:24 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:39 +0200

bus/arm-cci: Convert to hotplug statemachine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andrzej Hajda <a.hajda@samsung.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Olof Johansson <olof@lixom.net>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Punit Agrawal <punit.agrawal@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Suzuki K. Poulose <suzuki.poulose@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.679142601@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/bus/arm-cci.c      | 53 ++++++++++++++++++++--------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 24 insertions(+), 30 deletions(-)

diff --git a/drivers/bus/arm-cci.c b/drivers/bus/arm-cci.c
index a49b283..5755907f 100644
--- a/drivers/bus/arm-cci.c
+++ b/drivers/bus/arm-cci.c
@@ -144,12 +144,15 @@ struct cci_pmu {
 	int num_cntrs;
 	atomic_t active_events;
 	struct mutex reserve_mutex;
-	struct notifier_block cpu_nb;
+	struct list_head entry;
 	cpumask_t cpus;
 };
 
 #define to_cci_pmu(c)	(container_of(c, struct cci_pmu, pmu))
 
+static DEFINE_MUTEX(cci_pmu_mutex);
+static LIST_HEAD(cci_pmu_list);
+
 enum cci_models {
 #ifdef CONFIG_ARM_CCI400_PMU
 	CCI400_R0,
@@ -1503,31 +1506,26 @@ static int cci_pmu_init(struct cci_pmu *cci_pmu, struct platform_device *pdev)
 	return perf_pmu_register(&cci_pmu->pmu, name, -1);
 }
 
-static int cci_pmu_cpu_notifier(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+static int cci_pmu_offline_cpu(unsigned int cpu)
 {
-	struct cci_pmu *cci_pmu = container_of(self,
-					struct cci_pmu, cpu_nb);
-	unsigned int cpu = (long)hcpu;
+	struct cci_pmu *cci_pmu;
 	unsigned int target;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
+	mutex_lock(&cci_pmu_mutex);
+	list_for_each_entry(cci_pmu, &cci_pmu_list, entry) {
 		if (!cpumask_test_and_clear_cpu(cpu, &cci_pmu->cpus))
-			break;
+			continue;
 		target = cpumask_any_but(cpu_online_mask, cpu);
-		if (target >= nr_cpu_ids) // UP, last CPU
-			break;
+		if (target >= nr_cpu_ids)
+			continue;
 		/*
 		 * TODO: migrate context once core races on event->ctx have
 		 * been fixed.
 		 */
 		cpumask_set_cpu(target, &cci_pmu->cpus);
-	default:
-		break;
 	}
-
-	return NOTIFY_OK;
+	mutex_unlock(&cci_pmu_mutex);
+	return 0;
 }
 
 static struct cci_pmu_model cci_pmu_models[] = {
@@ -1766,24 +1764,13 @@ static int cci_pmu_probe(struct platform_device *pdev)
 	atomic_set(&cci_pmu->active_events, 0);
 	cpumask_set_cpu(smp_processor_id(), &cci_pmu->cpus);
 
-	cci_pmu->cpu_nb = (struct notifier_block) {
-		.notifier_call	= cci_pmu_cpu_notifier,
-		/*
-		 * to migrate uncore events, our notifier should be executed
-		 * before perf core's notifier.
-		 */
-		.priority	= CPU_PRI_PERF + 1,
-	};
-
-	ret = register_cpu_notifier(&cci_pmu->cpu_nb);
+	ret = cci_pmu_init(cci_pmu, pdev);
 	if (ret)
 		return ret;
 
-	ret = cci_pmu_init(cci_pmu, pdev);
-	if (ret) {
-		unregister_cpu_notifier(&cci_pmu->cpu_nb);
-		return ret;
-	}
+	mutex_lock(&cci_pmu_mutex);
+	list_add(&cci_pmu->entry, &cci_pmu_list);
+	mutex_unlock(&cci_pmu_mutex);
 
 	pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
 	return 0;
@@ -1817,6 +1804,12 @@ static int __init cci_platform_init(void)
 {
 	int ret;
 
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+					"AP_PERF_ARM_CCI_ONLINE", NULL,
+					cci_pmu_offline_cpu);
+	if (ret)
+		return ret;
+
 	ret = platform_driver_register(&cci_pmu_driver);
 	if (ret)
 		return ret;
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 16ab417..0f2f727 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -45,6 +45,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
 	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_PERF_S390_SF_ONLINE,
+	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 23/67] bus/arm-ccn: Convert to hotplug statemachine

[Anna-Maria Gleixner]

[-- Attachment #1: 0022-bus-arm-ccn-Convert-to-hotplug-statemachine.patch --]
[-- Type: text/plain, Size: 5102 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/bus/arm-ccn.c      |   57 ++++++++++++++++++++++-----------------------
 include/linux/cpuhotplug.h |    1 
 2 files changed, 29 insertions(+), 29 deletions(-)

--- a/drivers/bus/arm-ccn.c
+++ b/drivers/bus/arm-ccn.c
@@ -167,7 +167,7 @@ struct arm_ccn_dt {
 	struct hrtimer hrtimer;
 
 	cpumask_t cpu;
-	struct notifier_block cpu_nb;
+	struct list_head entry;
 
 	struct pmu pmu;
 };
@@ -189,6 +189,8 @@ struct arm_ccn {
 	struct arm_ccn_dt dt;
 };
 
+static DEFINE_MUTEX(arm_ccn_mutex);
+static LIST_HEAD(arm_ccn_list);
 
 static int arm_ccn_node_to_xp(int node)
 {
@@ -1171,30 +1173,27 @@ static enum hrtimer_restart arm_ccn_pmu_
 }
 
 
-static int arm_ccn_pmu_cpu_notifier(struct notifier_block *nb,
-		unsigned long action, void *hcpu)
+static int arm_ccn_pmu_offline_cpu(unsigned int cpu)
 {
-	struct arm_ccn_dt *dt = container_of(nb, struct arm_ccn_dt, cpu_nb);
-	struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
-	unsigned int cpu = (long)hcpu; /* for (long) see kernel/cpu.c */
+	struct arm_ccn_dt *dt;
 	unsigned int target;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
+	mutex_lock(&arm_ccn_mutex);
+	list_for_each_entry(dt, &arm_ccn_list, entry) {
+		struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
+
 		if (!cpumask_test_and_clear_cpu(cpu, &dt->cpu))
-			break;
+			continue;
 		target = cpumask_any_but(cpu_online_mask, cpu);
 		if (target >= nr_cpu_ids)
-			break;
+			continue;
 		perf_pmu_migrate_context(&dt->pmu, cpu, target);
 		cpumask_set_cpu(target, &dt->cpu);
 		if (ccn->irq)
 			WARN_ON(irq_set_affinity_hint(ccn->irq, &dt->cpu) != 0);
-	default:
-		break;
 	}
-
-	return NOTIFY_OK;
+	mutex_unlock(&arm_ccn_mutex);
+	return 0;
 }
 
 
@@ -1266,16 +1265,6 @@ static int arm_ccn_pmu_init(struct arm_c
 	/* Pick one CPU which we will use to collect data from CCN... */
 	cpumask_set_cpu(smp_processor_id(), &ccn->dt.cpu);
 
-	/*
-	 * ... and change the selection when it goes offline. Priority is
-	 * picked to have a chance to migrate events before perf is notified.
-	 */
-	ccn->dt.cpu_nb.notifier_call = arm_ccn_pmu_cpu_notifier;
-	ccn->dt.cpu_nb.priority = CPU_PRI_PERF + 1,
-	err = register_cpu_notifier(&ccn->dt.cpu_nb);
-	if (err)
-		goto error_cpu_notifier;
-
 	/* Also make sure that the overflow interrupt is handled by this CPU */
 	if (ccn->irq) {
 		err = irq_set_affinity_hint(ccn->irq, &ccn->dt.cpu);
@@ -1289,12 +1278,13 @@ static int arm_ccn_pmu_init(struct arm_c
 	if (err)
 		goto error_pmu_register;
 
+	mutex_lock(&arm_ccn_mutex);
+	list_add(&ccn->dt.entry, &arm_ccn_list);
+	mutex_unlock(&arm_ccn_mutex);
 	return 0;
 
 error_pmu_register:
 error_set_affinity:
-	unregister_cpu_notifier(&ccn->dt.cpu_nb);
-error_cpu_notifier:
 	ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id);
 	for (i = 0; i < ccn->num_xps; i++)
 		writel(0, ccn->xp[i].base + CCN_XP_DT_CONTROL);
@@ -1306,9 +1296,12 @@ static void arm_ccn_pmu_cleanup(struct a
 {
 	int i;
 
+	mutex_lock(&arm_ccn_mutex);
+	list_del(&ccn->dt.entry);
+	mutex_unlock(&arm_ccn_mutex);
+
 	if (ccn->irq)
 		irq_set_affinity_hint(ccn->irq, NULL);
-	unregister_cpu_notifier(&ccn->dt.cpu_nb);
 	for (i = 0; i < ccn->num_xps; i++)
 		writel(0, ccn->xp[i].base + CCN_XP_DT_CONTROL);
 	writel(0, ccn->dt.base + CCN_DT_PMCR);
@@ -1316,7 +1309,6 @@ static void arm_ccn_pmu_cleanup(struct a
 	ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id);
 }
 
-
 static int arm_ccn_for_each_valid_region(struct arm_ccn *ccn,
 		int (*callback)(struct arm_ccn *ccn, int region,
 		void __iomem *base, u32 type, u32 id))
@@ -1533,7 +1525,13 @@ static struct platform_driver arm_ccn_dr
 
 static int __init arm_ccn_init(void)
 {
-	int i;
+	int i, ret;
+
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					"AP_PERF_ARM_CCN_ONLINE", NULL,
+					arm_ccn_pmu_offline_cpu);
+	if (ret)
+		return ret;
 
 	for (i = 0; i < ARRAY_SIZE(arm_ccn_pmu_events); i++)
 		arm_ccn_pmu_events_attrs[i] = &arm_ccn_pmu_events[i].attr.attr;
@@ -1543,6 +1541,7 @@ static int __init arm_ccn_init(void)
 
 static void __exit arm_ccn_exit(void)
 {
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE);
 	platform_driver_unregister(&arm_ccn_driver);
 }
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -46,6 +46,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
+	CPUHP_AP_PERF_ARM_CCN_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

Re: [patch V2 23/67] bus/arm-ccn: Convert to hotplug statemachine

[Pawel Moll]

Dnia 2016-07-13, Wed o godzinie 17:16 +0000, Anna-Maria Gleixner pisze:
> From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> 
> Install the callbacks via the state machine and let the core invoke
> the callbacks on the already online CPUs.
> 
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Arnd Bergmann <arnd@arndb.de>
> Cc: Dan Carpenter <dan.carpenter@oracle.com>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Pawel Moll <pawel.moll@arm.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>

> @@ -1316,7 +1309,6 @@ static void arm_ccn_pmu_cleanup(struct a
>  	ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id);
>  }
>  
> -
>  static int arm_ccn_for_each_valid_region(struct arm_ccn *ccn,
>  		int (*callback)(struct arm_ccn *ccn, int region,
>  		void __iomem *base, u32 type, u32 id))

This empty line doesn't impact hotplug, does it? ;-)

With this chunk dropped

Acked-by: Pawel Moll <pawel.moll@arm.com>

Cheers!

Pawel

[tip:smp/hotplug] bus/arm-ccn: Convert to hotplug statemachine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  fdc15a36d84eeb52f58b907f03fd8f77f5648235
Gitweb:     http://git.kernel.org/tip/fdc15a36d84eeb52f58b907f03fd8f77f5648235
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:25 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:40 +0200

bus/arm-ccn: Convert to hotplug statemachine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.768498577@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/bus/arm-ccn.c      | 57 +++++++++++++++++++++++-----------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 29 insertions(+), 29 deletions(-)

diff --git a/drivers/bus/arm-ccn.c b/drivers/bus/arm-ccn.c
index acc3eb5..97a9185 100644
--- a/drivers/bus/arm-ccn.c
+++ b/drivers/bus/arm-ccn.c
@@ -167,7 +167,7 @@ struct arm_ccn_dt {
 	struct hrtimer hrtimer;
 
 	cpumask_t cpu;
-	struct notifier_block cpu_nb;
+	struct list_head entry;
 
 	struct pmu pmu;
 };
@@ -189,6 +189,8 @@ struct arm_ccn {
 	struct arm_ccn_dt dt;
 };
 
+static DEFINE_MUTEX(arm_ccn_mutex);
+static LIST_HEAD(arm_ccn_list);
 
 static int arm_ccn_node_to_xp(int node)
 {
@@ -1171,30 +1173,27 @@ static enum hrtimer_restart arm_ccn_pmu_timer_handler(struct hrtimer *hrtimer)
 }
 
 
-static int arm_ccn_pmu_cpu_notifier(struct notifier_block *nb,
-		unsigned long action, void *hcpu)
+static int arm_ccn_pmu_offline_cpu(unsigned int cpu)
 {
-	struct arm_ccn_dt *dt = container_of(nb, struct arm_ccn_dt, cpu_nb);
-	struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
-	unsigned int cpu = (long)hcpu; /* for (long) see kernel/cpu.c */
+	struct arm_ccn_dt *dt;
 	unsigned int target;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
+	mutex_lock(&arm_ccn_mutex);
+	list_for_each_entry(dt, &arm_ccn_list, entry) {
+		struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
+
 		if (!cpumask_test_and_clear_cpu(cpu, &dt->cpu))
-			break;
+			continue;
 		target = cpumask_any_but(cpu_online_mask, cpu);
 		if (target >= nr_cpu_ids)
-			break;
+			continue;
 		perf_pmu_migrate_context(&dt->pmu, cpu, target);
 		cpumask_set_cpu(target, &dt->cpu);
 		if (ccn->irq)
 			WARN_ON(irq_set_affinity_hint(ccn->irq, &dt->cpu) != 0);
-	default:
-		break;
 	}
-
-	return NOTIFY_OK;
+	mutex_unlock(&arm_ccn_mutex);
+	return 0;
 }
 
 
@@ -1266,16 +1265,6 @@ static int arm_ccn_pmu_init(struct arm_ccn *ccn)
 	/* Pick one CPU which we will use to collect data from CCN... */
 	cpumask_set_cpu(smp_processor_id(), &ccn->dt.cpu);
 
-	/*
-	 * ... and change the selection when it goes offline. Priority is
-	 * picked to have a chance to migrate events before perf is notified.
-	 */
-	ccn->dt.cpu_nb.notifier_call = arm_ccn_pmu_cpu_notifier;
-	ccn->dt.cpu_nb.priority = CPU_PRI_PERF + 1,
-	err = register_cpu_notifier(&ccn->dt.cpu_nb);
-	if (err)
-		goto error_cpu_notifier;
-
 	/* Also make sure that the overflow interrupt is handled by this CPU */
 	if (ccn->irq) {
 		err = irq_set_affinity_hint(ccn->irq, &ccn->dt.cpu);
@@ -1289,12 +1278,13 @@ static int arm_ccn_pmu_init(struct arm_ccn *ccn)
 	if (err)
 		goto error_pmu_register;
 
+	mutex_lock(&arm_ccn_mutex);
+	list_add(&ccn->dt.entry, &arm_ccn_list);
+	mutex_unlock(&arm_ccn_mutex);
 	return 0;
 
 error_pmu_register:
 error_set_affinity:
-	unregister_cpu_notifier(&ccn->dt.cpu_nb);
-error_cpu_notifier:
 	ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id);
 	for (i = 0; i < ccn->num_xps; i++)
 		writel(0, ccn->xp[i].base + CCN_XP_DT_CONTROL);
@@ -1306,9 +1296,12 @@ static void arm_ccn_pmu_cleanup(struct arm_ccn *ccn)
 {
 	int i;
 
+	mutex_lock(&arm_ccn_mutex);
+	list_del(&ccn->dt.entry);
+	mutex_unlock(&arm_ccn_mutex);
+
 	if (ccn->irq)
 		irq_set_affinity_hint(ccn->irq, NULL);
-	unregister_cpu_notifier(&ccn->dt.cpu_nb);
 	for (i = 0; i < ccn->num_xps; i++)
 		writel(0, ccn->xp[i].base + CCN_XP_DT_CONTROL);
 	writel(0, ccn->dt.base + CCN_DT_PMCR);
@@ -1316,7 +1309,6 @@ static void arm_ccn_pmu_cleanup(struct arm_ccn *ccn)
 	ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id);
 }
 
-
 static int arm_ccn_for_each_valid_region(struct arm_ccn *ccn,
 		int (*callback)(struct arm_ccn *ccn, int region,
 		void __iomem *base, u32 type, u32 id))
@@ -1533,7 +1525,13 @@ static struct platform_driver arm_ccn_driver = {
 
 static int __init arm_ccn_init(void)
 {
-	int i;
+	int i, ret;
+
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					"AP_PERF_ARM_CCN_ONLINE", NULL,
+					arm_ccn_pmu_offline_cpu);
+	if (ret)
+		return ret;
 
 	for (i = 0; i < ARRAY_SIZE(arm_ccn_pmu_events); i++)
 		arm_ccn_pmu_events_attrs[i] = &arm_ccn_pmu_events[i].attr.attr;
@@ -1543,6 +1541,7 @@ static int __init arm_ccn_init(void)
 
 static void __exit arm_ccn_exit(void)
 {
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE);
 	platform_driver_unregister(&arm_ccn_driver);
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 0f2f727..2ee3c0e 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -46,6 +46,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
+	CPUHP_AP_PERF_ARM_CCN_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,

[patch V2 24/67] xtensa/perf: Convert the hotplug notifier to state machine callbacks

[Anna-Maria Gleixner]

[-- Attachment #1: 0023-xtensa-perf-Convert-the-hotplug-notifier-to-state-ma.patch --]
[-- Type: text/plain, Size: 2292 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Chris Zankel <chris@zankel.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-xtensa@linux-xtensa.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/xtensa/kernel/perf_event.c |   26 +++++++++-----------------
 include/linux/cpuhotplug.h      |    1 +
 2 files changed, 10 insertions(+), 17 deletions(-)

--- a/arch/xtensa/kernel/perf_event.c
+++ b/arch/xtensa/kernel/perf_event.c
@@ -404,7 +404,7 @@ static struct pmu xtensa_pmu = {
 	.read = xtensa_pmu_read,
 };
 
-static void xtensa_pmu_setup(void)
+static int xtensa_pmu_setup(int cpu)
 {
 	unsigned i;
 
@@ -413,21 +413,7 @@ static void xtensa_pmu_setup(void)
 		set_er(0, XTENSA_PMU_PMCTRL(i));
 		set_er(get_er(XTENSA_PMU_PMSTAT(i)), XTENSA_PMU_PMSTAT(i));
 	}
-}
-
-static int xtensa_pmu_notifier(struct notifier_block *self,
-			       unsigned long action, void *data)
-{
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		xtensa_pmu_setup();
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 static int __init xtensa_pmu_init(void)
@@ -435,7 +421,13 @@ static int __init xtensa_pmu_init(void)
 	int ret;
 	int irq = irq_create_mapping(NULL, XCHAL_PROFILING_INTERRUPT);
 
-	perf_cpu_notifier(xtensa_pmu_notifier);
+	ret = cpuhp_setup_state(CPUHP_AP_PERF_XTENSA_STARTING,
+				"AP_PERF_XTENSA_STARTING", xtensa_pmu_setup,
+				NULL);
+	if (ret) {
+		pr_err("xtensa_pmu: failed to register CPU-hotplug.\n");
+		return ret;
+	}
 #if XTENSA_FAKE_NMI
 	enable_irq(irq);
 #else
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -30,6 +30,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
+	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[tip:smp/hotplug] xtensa/perf: Convert the hotplug notifier to state machine callbacks

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  25a77b55e74c46c79f838a97e782a45c33588ca6
Gitweb:     http://git.kernel.org/tip/25a77b55e74c46c79f838a97e782a45c33588ca6
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:26 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:41 +0200

xtensa/perf: Convert the hotplug notifier to state machine callbacks

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Chris Zankel <chris@zankel.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-xtensa@linux-xtensa.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.852575891@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/xtensa/kernel/perf_event.c | 26 +++++++++-----------------
 include/linux/cpuhotplug.h      |  1 +
 2 files changed, 10 insertions(+), 17 deletions(-)

diff --git a/arch/xtensa/kernel/perf_event.c b/arch/xtensa/kernel/perf_event.c
index ef90479..0fecc8a 100644
--- a/arch/xtensa/kernel/perf_event.c
+++ b/arch/xtensa/kernel/perf_event.c
@@ -404,7 +404,7 @@ static struct pmu xtensa_pmu = {
 	.read = xtensa_pmu_read,
 };
 
-static void xtensa_pmu_setup(void)
+static int xtensa_pmu_setup(int cpu)
 {
 	unsigned i;
 
@@ -413,21 +413,7 @@ static void xtensa_pmu_setup(void)
 		set_er(0, XTENSA_PMU_PMCTRL(i));
 		set_er(get_er(XTENSA_PMU_PMSTAT(i)), XTENSA_PMU_PMSTAT(i));
 	}
-}
-
-static int xtensa_pmu_notifier(struct notifier_block *self,
-			       unsigned long action, void *data)
-{
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		xtensa_pmu_setup();
-		break;
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
+	return 0;
 }
 
 static int __init xtensa_pmu_init(void)
@@ -435,7 +421,13 @@ static int __init xtensa_pmu_init(void)
 	int ret;
 	int irq = irq_create_mapping(NULL, XCHAL_PROFILING_INTERRUPT);
 
-	perf_cpu_notifier(xtensa_pmu_notifier);
+	ret = cpuhp_setup_state(CPUHP_AP_PERF_XTENSA_STARTING,
+				"AP_PERF_XTENSA_STARTING", xtensa_pmu_setup,
+				NULL);
+	if (ret) {
+		pr_err("xtensa_pmu: failed to register CPU-hotplug.\n");
+		return ret;
+	}
 #if XTENSA_FAKE_NMI
 	enable_irq(irq);
 #else
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 2ee3c0e..f187c46 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -30,6 +30,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_AMD_IBS_STARTING,
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
+	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 25/67] perf/x86/amd/power: Change hotplug notifier to a symmetric structure

[Anna-Maria Gleixner]

[-- Attachment #1: 0024-perf-x86-amd-power-Change-hotplug-notifier-to-a-symm.patch --]
[-- Type: text/plain, Size: 1401 bytes --]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

To simplify the hotplug mechanism move the starting callback to
online. There is no functional requirement that the cpumask bit has to
be set in the starting callback.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
---
 arch/x86/events/amd/power.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/events/amd/power.c b/arch/x86/events/amd/power.c
index 55a3529..204df18 100644
--- a/arch/x86/events/amd/power.c
+++ b/arch/x86/events/amd/power.c
@@ -255,7 +255,7 @@ static void power_cpu_init(int cpu)
 	 * 1) If any CPU is set at cpu_mask in the same compute unit, do
 	 * nothing.
 	 * 2) If no CPU is set at cpu_mask in the same compute unit,
-	 * set current STARTING CPU.
+	 * set current ONLINE CPU.
 	 *
 	 * Note: if there is a CPU aside of the new one already in the
 	 * sibling mask, then it is also in cpu_mask.
@@ -272,7 +272,7 @@ power_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu
 
 	switch (action & ~CPU_TASKS_FROZEN) {
 	case CPU_DOWN_FAILED:
-	case CPU_STARTING:
+	case CPU_ONLINE:
 		power_cpu_init(cpu);
 		break;
 	case CPU_DOWN_PREPARE:
-- 
2.8.1

[tip:smp/hotplug] perf/x86/amd/power: Change hotplug notifier to a symmetric structure

[tip-bot for Anna-Maria Gleixner]

Commit-ID:  8381f6a0c02fdf457d1712cb82f2f0240f2f03cc
Gitweb:     http://git.kernel.org/tip/8381f6a0c02fdf457d1712cb82f2f0240f2f03cc
Author:     Anna-Maria Gleixner <anna-maria@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:27 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:41 +0200

perf/x86/amd/power: Change hotplug notifier to a symmetric structure

To simplify the hotplug mechanism move the starting callback to
online. There is no functional requirement that the cpumask bit has to
be set in the starting callback.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153334.944849172@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/amd/power.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/events/amd/power.c b/arch/x86/events/amd/power.c
index 55a3529..204df18 100644
--- a/arch/x86/events/amd/power.c
+++ b/arch/x86/events/amd/power.c
@@ -255,7 +255,7 @@ static void power_cpu_init(int cpu)
 	 * 1) If any CPU is set at cpu_mask in the same compute unit, do
 	 * nothing.
 	 * 2) If no CPU is set at cpu_mask in the same compute unit,
-	 * set current STARTING CPU.
+	 * set current ONLINE CPU.
 	 *
 	 * Note: if there is a CPU aside of the new one already in the
 	 * sibling mask, then it is also in cpu_mask.
@@ -272,7 +272,7 @@ power_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu
 
 	switch (action & ~CPU_TASKS_FROZEN) {
 	case CPU_DOWN_FAILED:
-	case CPU_STARTING:
+	case CPU_ONLINE:
 		power_cpu_init(cpu);
 		break;
 	case CPU_DOWN_PREPARE:

[patch V2 26/67] perf/x86/amd/power: Convert the hotplug notifier to state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0025-perf-x86-amd-power-Convert-the-hotplug-notifier-to-s.patch --]
[-- Type: text/plain, Size: 3906 bytes --]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
---
 arch/x86/events/amd/power.c | 58 +++++++++------------------------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 12 insertions(+), 47 deletions(-)

diff --git a/arch/x86/events/amd/power.c b/arch/x86/events/amd/power.c
index 204df18..9842270 100644
--- a/arch/x86/events/amd/power.c
+++ b/arch/x86/events/amd/power.c
@@ -228,12 +228,12 @@ static struct pmu pmu_class = {
 	.read		= pmu_event_read,
 };
 
-static void power_cpu_exit(int cpu)
+static int power_cpu_exit(unsigned int cpu)
 {
 	int target;
 
 	if (!cpumask_test_and_clear_cpu(cpu, &cpu_mask))
-		return;
+		return 0;
 
 	/*
 	 * Find a new CPU on the same compute unit, if was set in cpumask
@@ -245,9 +245,10 @@ static void power_cpu_exit(int cpu)
 		cpumask_set_cpu(target, &cpu_mask);
 		perf_pmu_migrate_context(&pmu_class, cpu, target);
 	}
+	return 0;
 }
 
-static void power_cpu_init(int cpu)
+static int power_cpu_init(unsigned int cpu)
 {
 	int target;
 
@@ -263,33 +264,9 @@ static void power_cpu_init(int cpu)
 	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
 	if (target >= nr_cpumask_bits)
 		cpumask_set_cpu(cpu, &cpu_mask);
+	return 0;
 }
 
-static int
-power_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		power_cpu_init(cpu);
-		break;
-	case CPU_DOWN_PREPARE:
-		power_cpu_exit(cpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block power_cpu_notifier_nb = {
-	.notifier_call = power_cpu_notifier,
-	.priority = CPU_PRI_PERF,
-};
-
 static const struct x86_cpu_id cpu_match[] = {
 	{ .vendor = X86_VENDOR_AMD, .family = 0x15 },
 	{},
@@ -297,7 +274,7 @@ static const struct x86_cpu_id cpu_match[] = {
 
 static int __init amd_power_pmu_init(void)
 {
-	int cpu, target, ret;
+	int ret;
 
 	if (!x86_match_cpu(cpu_match))
 		return 0;
@@ -312,38 +289,25 @@ static int __init amd_power_pmu_init(void)
 		return -ENODEV;
 	}
 
-	cpu_notifier_register_begin();
 
-	/* Choose one online core of each compute unit. */
-	for_each_online_cpu(cpu) {
-		target = cpumask_first(topology_sibling_cpumask(cpu));
-		if (!cpumask_test_cpu(target, &cpu_mask))
-			cpumask_set_cpu(target, &cpu_mask);
-	}
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
+			  "AP_PERF_X86_AMD_POWER_ONLINE",
+			  power_cpu_init, power_cpu_exit);
 
 	ret = perf_pmu_register(&pmu_class, "power", -1);
 	if (WARN_ON(ret)) {
 		pr_warn("AMD Power PMU registration failed\n");
-		goto out;
+		return ret;
 	}
 
-	__register_cpu_notifier(&power_cpu_notifier_nb);
-
 	pr_info("AMD Power PMU detected\n");
-
-out:
-	cpu_notifier_register_done();
-
 	return ret;
 }
 module_init(amd_power_pmu_init);
 
 static void __exit amd_power_pmu_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&power_cpu_notifier_nb);
-	cpu_notifier_register_done();
-
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE);
 	perf_pmu_unregister(&pmu_class);
 }
 module_exit(amd_power_pmu_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f187c46..acfeda1 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -41,6 +41,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] perf/x86/amd/power: Convert the hotplug notifier to state machine

[tip-bot for Anna-Maria Gleixner]

Commit-ID:  c6a84daa3498f269e3b506757506f762b6f56080
Gitweb:     http://git.kernel.org/tip/c6a84daa3498f269e3b506757506f762b6f56080
Author:     Anna-Maria Gleixner <anna-maria@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:28 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:42 +0200

perf/x86/amd/power: Convert the hotplug notifier to state machine

Install the callbacks via the state machine.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Huang Rui <ray.huang@amd.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.027571056@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/amd/power.c | 58 +++++++++------------------------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 12 insertions(+), 47 deletions(-)

diff --git a/arch/x86/events/amd/power.c b/arch/x86/events/amd/power.c
index 204df18..9842270 100644
--- a/arch/x86/events/amd/power.c
+++ b/arch/x86/events/amd/power.c
@@ -228,12 +228,12 @@ static struct pmu pmu_class = {
 	.read		= pmu_event_read,
 };
 
-static void power_cpu_exit(int cpu)
+static int power_cpu_exit(unsigned int cpu)
 {
 	int target;
 
 	if (!cpumask_test_and_clear_cpu(cpu, &cpu_mask))
-		return;
+		return 0;
 
 	/*
 	 * Find a new CPU on the same compute unit, if was set in cpumask
@@ -245,9 +245,10 @@ static void power_cpu_exit(int cpu)
 		cpumask_set_cpu(target, &cpu_mask);
 		perf_pmu_migrate_context(&pmu_class, cpu, target);
 	}
+	return 0;
 }
 
-static void power_cpu_init(int cpu)
+static int power_cpu_init(unsigned int cpu)
 {
 	int target;
 
@@ -263,33 +264,9 @@ static void power_cpu_init(int cpu)
 	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
 	if (target >= nr_cpumask_bits)
 		cpumask_set_cpu(cpu, &cpu_mask);
+	return 0;
 }
 
-static int
-power_cpu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		power_cpu_init(cpu);
-		break;
-	case CPU_DOWN_PREPARE:
-		power_cpu_exit(cpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block power_cpu_notifier_nb = {
-	.notifier_call = power_cpu_notifier,
-	.priority = CPU_PRI_PERF,
-};
-
 static const struct x86_cpu_id cpu_match[] = {
 	{ .vendor = X86_VENDOR_AMD, .family = 0x15 },
 	{},
@@ -297,7 +274,7 @@ static const struct x86_cpu_id cpu_match[] = {
 
 static int __init amd_power_pmu_init(void)
 {
-	int cpu, target, ret;
+	int ret;
 
 	if (!x86_match_cpu(cpu_match))
 		return 0;
@@ -312,38 +289,25 @@ static int __init amd_power_pmu_init(void)
 		return -ENODEV;
 	}
 
-	cpu_notifier_register_begin();
 
-	/* Choose one online core of each compute unit. */
-	for_each_online_cpu(cpu) {
-		target = cpumask_first(topology_sibling_cpumask(cpu));
-		if (!cpumask_test_cpu(target, &cpu_mask))
-			cpumask_set_cpu(target, &cpu_mask);
-	}
+	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
+			  "AP_PERF_X86_AMD_POWER_ONLINE",
+			  power_cpu_init, power_cpu_exit);
 
 	ret = perf_pmu_register(&pmu_class, "power", -1);
 	if (WARN_ON(ret)) {
 		pr_warn("AMD Power PMU registration failed\n");
-		goto out;
+		return ret;
 	}
 
-	__register_cpu_notifier(&power_cpu_notifier_nb);
-
 	pr_info("AMD Power PMU detected\n");
-
-out:
-	cpu_notifier_register_done();
-
 	return ret;
 }
 module_init(amd_power_pmu_init);
 
 static void __exit amd_power_pmu_exit(void)
 {
-	cpu_notifier_register_begin();
-	__unregister_cpu_notifier(&power_cpu_notifier_nb);
-	cpu_notifier_register_done();
-
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE);
 	perf_pmu_unregister(&pmu_class);
 }
 module_exit(amd_power_pmu_exit);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f187c46..acfeda1 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -41,6 +41,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_ONLINE,
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
+	CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
 	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_X86_CQM_ONLINE,
 	CPUHP_AP_PERF_X86_CSTATE_ONLINE,

[patch V2 27/67] perf/core: Remove perf CPU notifier code

[Anna-Maria Gleixner]

[-- Attachment #1: 0026-perf-core-Remove-perf-CPU-notifier-code.patch --]
[-- Type: text/plain, Size: 2629 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

All users converted to state machine callbacks.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpu.h        |  2 --
 include/linux/perf_event.h | 35 -----------------------------------
 2 files changed, 37 deletions(-)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 21597dc..ca2dd86 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -59,8 +59,6 @@ struct notifier_block;
  * CPU notifier priorities.
  */
 enum {
-	CPU_PRI_PERF		= 20,
-
 	/* bring up workqueues before normal notifiers and down after */
 	CPU_PRI_WORKQUEUE_UP	= 5,
 	CPU_PRI_WORKQUEUE_DOWN	= -5,
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 9abeb69..ddd3dab 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1285,41 +1285,6 @@ static inline void perf_restore_debug_store(void)			{ }
 
 #define perf_output_put(handle, x) perf_output_copy((handle), &(x), sizeof(x))
 
-/*
- * This has to have a higher priority than migration_notifier in sched/core.c.
- */
-#define perf_cpu_notifier(fn)						\
-do {									\
-	static struct notifier_block fn##_nb =				\
-		{ .notifier_call = fn, .priority = CPU_PRI_PERF };	\
-	unsigned long cpu = smp_processor_id();				\
-	unsigned long flags;						\
-									\
-	cpu_notifier_register_begin();					\
-	fn(&fn##_nb, (unsigned long)CPU_UP_PREPARE,			\
-		(void *)(unsigned long)cpu);				\
-	local_irq_save(flags);						\
-	fn(&fn##_nb, (unsigned long)CPU_STARTING,			\
-		(void *)(unsigned long)cpu);				\
-	local_irq_restore(flags);					\
-	fn(&fn##_nb, (unsigned long)CPU_ONLINE,				\
-		(void *)(unsigned long)cpu);				\
-	__register_cpu_notifier(&fn##_nb);				\
-	cpu_notifier_register_done();					\
-} while (0)
-
-/*
- * Bare-bones version of perf_cpu_notifier(), which doesn't invoke the
- * callback for already online CPUs.
- */
-#define __perf_cpu_notifier(fn)						\
-do {									\
-	static struct notifier_block fn##_nb =				\
-		{ .notifier_call = fn, .priority = CPU_PRI_PERF };	\
-									\
-	__register_cpu_notifier(&fn##_nb);				\
-} while (0)
-
 struct perf_pmu_events_attr {
 	struct device_attribute attr;
 	u64 id;
-- 
2.8.1

[tip:smp/hotplug] perf/core: Remove perf CPU notifier code

[tip-bot for Thomas Gleixner]

Commit-ID:  89ab9cb16931873ec600a909b3a38436352e629a
Gitweb:     http://git.kernel.org/tip/89ab9cb16931873ec600a909b3a38436352e629a
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:28 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:42 +0200

perf/core: Remove perf CPU notifier code

All users converted to state machine callbacks.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.115333381@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpu.h        |  2 --
 include/linux/perf_event.h | 35 -----------------------------------
 2 files changed, 37 deletions(-)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 21597dc..ca2dd86 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -59,8 +59,6 @@ struct notifier_block;
  * CPU notifier priorities.
  */
 enum {
-	CPU_PRI_PERF		= 20,
-
 	/* bring up workqueues before normal notifiers and down after */
 	CPU_PRI_WORKQUEUE_UP	= 5,
 	CPU_PRI_WORKQUEUE_DOWN	= -5,
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 9abeb69..ddd3dab 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1285,41 +1285,6 @@ static inline void perf_restore_debug_store(void)			{ }
 
 #define perf_output_put(handle, x) perf_output_copy((handle), &(x), sizeof(x))
 
-/*
- * This has to have a higher priority than migration_notifier in sched/core.c.
- */
-#define perf_cpu_notifier(fn)						\
-do {									\
-	static struct notifier_block fn##_nb =				\
-		{ .notifier_call = fn, .priority = CPU_PRI_PERF };	\
-	unsigned long cpu = smp_processor_id();				\
-	unsigned long flags;						\
-									\
-	cpu_notifier_register_begin();					\
-	fn(&fn##_nb, (unsigned long)CPU_UP_PREPARE,			\
-		(void *)(unsigned long)cpu);				\
-	local_irq_save(flags);						\
-	fn(&fn##_nb, (unsigned long)CPU_STARTING,			\
-		(void *)(unsigned long)cpu);				\
-	local_irq_restore(flags);					\
-	fn(&fn##_nb, (unsigned long)CPU_ONLINE,				\
-		(void *)(unsigned long)cpu);				\
-	__register_cpu_notifier(&fn##_nb);				\
-	cpu_notifier_register_done();					\
-} while (0)
-
-/*
- * Bare-bones version of perf_cpu_notifier(), which doesn't invoke the
- * callback for already online CPUs.
- */
-#define __perf_cpu_notifier(fn)						\
-do {									\
-	static struct notifier_block fn##_nb =				\
-		{ .notifier_call = fn, .priority = CPU_PRI_PERF };	\
-									\
-	__register_cpu_notifier(&fn##_nb);				\
-} while (0)
-
 struct perf_pmu_events_attr {
 	struct device_attribute attr;
 	u64 id;

[patch V2 28/67] workqueue: Convert to state machine callbacks

[Anna-Maria Gleixner]

[-- Attachment #1: 0027-workqueue-Convert-to-state-machine-callbacks.patch --]
[-- Type: text/plain, Size: 7062 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Get rid of the prio ordering of the separate notifiers and use a proper state
callback pair.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpu.h        |    9 ---
 include/linux/cpuhotplug.h |    2 
 include/linux/workqueue.h  |    6 ++
 kernel/cpu.c               |   10 ++++
 kernel/workqueue.c         |  108 +++++++++++++++++----------------------------
 5 files changed, 61 insertions(+), 74 deletions(-)

--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -55,15 +55,6 @@ extern ssize_t arch_cpu_release(const ch
 #endif
 struct notifier_block;
 
-/*
- * CPU notifier priorities.
- */
-enum {
-	/* bring up workqueues before normal notifiers and down after */
-	CPU_PRI_WORKQUEUE_UP	= 5,
-	CPU_PRI_WORKQUEUE_DOWN	= -5,
-};
-
 #define CPU_ONLINE		0x0002 /* CPU (unsigned)v is up */
 #define CPU_UP_PREPARE		0x0003 /* CPU (unsigned)v coming up */
 #define CPU_UP_CANCELED		0x0004 /* CPU (unsigned)v NOT coming up */
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,7 @@ enum cpuhp_state {
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
+	CPUHP_WORKQUEUE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -49,6 +50,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_PERF_ARM_CCN_ONLINE,
+	CPUHP_AP_WORKQUEUE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
--- a/include/linux/workqueue.h
+++ b/include/linux/workqueue.h
@@ -625,4 +625,10 @@ void wq_watchdog_touch(int cpu);
 static inline void wq_watchdog_touch(int cpu) { }
 #endif	/* CONFIG_WQ_WATCHDOG */
 
+#ifdef CONFIG_SMP
+int workqueue_prepare_cpu(unsigned int cpu);
+int workqueue_online_cpu(unsigned int cpu);
+int workqueue_offline_cpu(unsigned int cpu);
+#endif
+
 #endif
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1185,6 +1185,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.startup = perf_event_init_cpu,
 		.teardown = perf_event_exit_cpu,
 	},
+	[CPUHP_WORKQUEUE_PREP] = {
+		.name = "workqueue prepare",
+		.startup = workqueue_prepare_cpu,
+		.teardown = NULL,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1265,6 +1270,11 @@ static struct cpuhp_step cpuhp_ap_states
 		.startup = perf_event_init_cpu,
 		.teardown = perf_event_exit_cpu,
 	},
+	[CPUHP_AP_WORKQUEUE_ONLINE] = {
+		.name = "workqueue online",
+		.startup = workqueue_online_cpu,
+		.teardown = workqueue_offline_cpu,
+	},
 
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -4611,84 +4611,65 @@ static void restore_unbound_workers_cpum
 						  pool->attrs->cpumask) < 0);
 }
 
-/*
- * Workqueues should be brought up before normal priority CPU notifiers.
- * This will be registered high priority CPU notifier.
- */
-static int workqueue_cpu_up_callback(struct notifier_block *nfb,
-					       unsigned long action,
-					       void *hcpu)
+int workqueue_prepare_cpu(unsigned int cpu)
+{
+	struct worker_pool *pool;
+
+	for_each_cpu_worker_pool(pool, cpu) {
+		if (pool->nr_workers)
+			continue;
+		if (!create_worker(pool))
+			return -ENOMEM;
+	}
+	return 0;
+}
+
+int workqueue_online_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
 	struct worker_pool *pool;
 	struct workqueue_struct *wq;
 	int pi;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		for_each_cpu_worker_pool(pool, cpu) {
-			if (pool->nr_workers)
-				continue;
-			if (!create_worker(pool))
-				return NOTIFY_BAD;
-		}
-		break;
-
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		mutex_lock(&wq_pool_mutex);
+	mutex_lock(&wq_pool_mutex);
 
-		for_each_pool(pool, pi) {
-			mutex_lock(&pool->attach_mutex);
+	for_each_pool(pool, pi) {
+		mutex_lock(&pool->attach_mutex);
 
-			if (pool->cpu == cpu)
-				rebind_workers(pool);
-			else if (pool->cpu < 0)
-				restore_unbound_workers_cpumask(pool, cpu);
+		if (pool->cpu == cpu)
+			rebind_workers(pool);
+		else if (pool->cpu < 0)
+			restore_unbound_workers_cpumask(pool, cpu);
 
-			mutex_unlock(&pool->attach_mutex);
-		}
+		mutex_unlock(&pool->attach_mutex);
+	}
 
-		/* update NUMA affinity of unbound workqueues */
-		list_for_each_entry(wq, &workqueues, list)
-			wq_update_unbound_numa(wq, cpu, true);
+	/* update NUMA affinity of unbound workqueues */
+	list_for_each_entry(wq, &workqueues, list)
+		wq_update_unbound_numa(wq, cpu, true);
 
-		mutex_unlock(&wq_pool_mutex);
-		break;
-	}
-	return NOTIFY_OK;
+	mutex_unlock(&wq_pool_mutex);
+	return 0;
 }
 
-/*
- * Workqueues should be brought down after normal priority CPU notifiers.
- * This will be registered as low priority CPU notifier.
- */
-static int workqueue_cpu_down_callback(struct notifier_block *nfb,
-						 unsigned long action,
-						 void *hcpu)
+int workqueue_offline_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
 	struct work_struct unbind_work;
 	struct workqueue_struct *wq;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
-		/* unbinding per-cpu workers should happen on the local CPU */
-		INIT_WORK_ONSTACK(&unbind_work, wq_unbind_fn);
-		queue_work_on(cpu, system_highpri_wq, &unbind_work);
-
-		/* update NUMA affinity of unbound workqueues */
-		mutex_lock(&wq_pool_mutex);
-		list_for_each_entry(wq, &workqueues, list)
-			wq_update_unbound_numa(wq, cpu, false);
-		mutex_unlock(&wq_pool_mutex);
-
-		/* wait for per-cpu unbinding to finish */
-		flush_work(&unbind_work);
-		destroy_work_on_stack(&unbind_work);
-		break;
-	}
-	return NOTIFY_OK;
+	/* unbinding per-cpu workers should happen on the local CPU */
+	INIT_WORK_ONSTACK(&unbind_work, wq_unbind_fn);
+	queue_work_on(cpu, system_highpri_wq, &unbind_work);
+
+	/* update NUMA affinity of unbound workqueues */
+	mutex_lock(&wq_pool_mutex);
+	list_for_each_entry(wq, &workqueues, list)
+		wq_update_unbound_numa(wq, cpu, false);
+	mutex_unlock(&wq_pool_mutex);
+
+	/* wait for per-cpu unbinding to finish */
+	flush_work(&unbind_work);
+	destroy_work_on_stack(&unbind_work);
+	return 0;
 }
 
 #ifdef CONFIG_SMP
@@ -5490,9 +5471,6 @@ static int __init init_workqueues(void)
 
 	pwq_cache = KMEM_CACHE(pool_workqueue, SLAB_PANIC);
 
-	cpu_notifier(workqueue_cpu_up_callback, CPU_PRI_WORKQUEUE_UP);
-	hotcpu_notifier(workqueue_cpu_down_callback, CPU_PRI_WORKQUEUE_DOWN);
-
 	wq_numa_init();
 
 	/* initialize CPU pools */

[tip:smp/hotplug] workqueue: Convert to state machine callbacks

[tip-bot for Thomas Gleixner]

Commit-ID:  7ee681b25284782ecf380bf5ccf55f13c52fd0ce
Gitweb:     http://git.kernel.org/tip/7ee681b25284782ecf380bf5ccf55f13c52fd0ce
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:29 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:43 +0200

workqueue: Convert to state machine callbacks

Get rid of the prio ordering of the separate notifiers and use a proper state
callback pair.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.197083890@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpu.h        |   9 ----
 include/linux/cpuhotplug.h |   2 +
 include/linux/workqueue.h  |   6 +++
 kernel/cpu.c               |  10 +++++
 kernel/workqueue.c         | 108 ++++++++++++++++++---------------------------
 5 files changed, 61 insertions(+), 74 deletions(-)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index ca2dd86..797d9c8 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -55,15 +55,6 @@ extern ssize_t arch_cpu_release(const char *, size_t);
 #endif
 struct notifier_block;
 
-/*
- * CPU notifier priorities.
- */
-enum {
-	/* bring up workqueues before normal notifiers and down after */
-	CPU_PRI_WORKQUEUE_UP	= 5,
-	CPU_PRI_WORKQUEUE_DOWN	= -5,
-};
-
 #define CPU_ONLINE		0x0002 /* CPU (unsigned)v is up */
 #define CPU_UP_PREPARE		0x0003 /* CPU (unsigned)v coming up */
 #define CPU_UP_CANCELED		0x0004 /* CPU (unsigned)v NOT coming up */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index acfeda1..60557a9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,7 @@ enum cpuhp_state {
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
+	CPUHP_WORKQUEUE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
@@ -49,6 +50,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_PERF_ARM_CCN_ONLINE,
+	CPUHP_AP_WORKQUEUE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h
index ca73c50..26cc1df 100644
--- a/include/linux/workqueue.h
+++ b/include/linux/workqueue.h
@@ -625,4 +625,10 @@ void wq_watchdog_touch(int cpu);
 static inline void wq_watchdog_touch(int cpu) { }
 #endif	/* CONFIG_WQ_WATCHDOG */
 
+#ifdef CONFIG_SMP
+int workqueue_prepare_cpu(unsigned int cpu);
+int workqueue_online_cpu(unsigned int cpu);
+int workqueue_offline_cpu(unsigned int cpu);
+#endif
+
 #endif
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 3705d90..af53f82 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1185,6 +1185,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = perf_event_init_cpu,
 		.teardown = perf_event_exit_cpu,
 	},
+	[CPUHP_WORKQUEUE_PREP] = {
+		.name = "workqueue prepare",
+		.startup = workqueue_prepare_cpu,
+		.teardown = NULL,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1267,6 +1272,11 @@ static struct cpuhp_step cpuhp_ap_states[] = {
 		.startup = perf_event_init_cpu,
 		.teardown = perf_event_exit_cpu,
 	},
+	[CPUHP_AP_WORKQUEUE_ONLINE] = {
+		.name = "workqueue online",
+		.startup = workqueue_online_cpu,
+		.teardown = workqueue_offline_cpu,
+	},
 
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index e1c0e99..c9dd5fb 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -4611,84 +4611,65 @@ static void restore_unbound_workers_cpumask(struct worker_pool *pool, int cpu)
 						  pool->attrs->cpumask) < 0);
 }
 
-/*
- * Workqueues should be brought up before normal priority CPU notifiers.
- * This will be registered high priority CPU notifier.
- */
-static int workqueue_cpu_up_callback(struct notifier_block *nfb,
-					       unsigned long action,
-					       void *hcpu)
+int workqueue_prepare_cpu(unsigned int cpu)
+{
+	struct worker_pool *pool;
+
+	for_each_cpu_worker_pool(pool, cpu) {
+		if (pool->nr_workers)
+			continue;
+		if (!create_worker(pool))
+			return -ENOMEM;
+	}
+	return 0;
+}
+
+int workqueue_online_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
 	struct worker_pool *pool;
 	struct workqueue_struct *wq;
 	int pi;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_UP_PREPARE:
-		for_each_cpu_worker_pool(pool, cpu) {
-			if (pool->nr_workers)
-				continue;
-			if (!create_worker(pool))
-				return NOTIFY_BAD;
-		}
-		break;
-
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
-		mutex_lock(&wq_pool_mutex);
+	mutex_lock(&wq_pool_mutex);
 
-		for_each_pool(pool, pi) {
-			mutex_lock(&pool->attach_mutex);
+	for_each_pool(pool, pi) {
+		mutex_lock(&pool->attach_mutex);
 
-			if (pool->cpu == cpu)
-				rebind_workers(pool);
-			else if (pool->cpu < 0)
-				restore_unbound_workers_cpumask(pool, cpu);
+		if (pool->cpu == cpu)
+			rebind_workers(pool);
+		else if (pool->cpu < 0)
+			restore_unbound_workers_cpumask(pool, cpu);
 
-			mutex_unlock(&pool->attach_mutex);
-		}
+		mutex_unlock(&pool->attach_mutex);
+	}
 
-		/* update NUMA affinity of unbound workqueues */
-		list_for_each_entry(wq, &workqueues, list)
-			wq_update_unbound_numa(wq, cpu, true);
+	/* update NUMA affinity of unbound workqueues */
+	list_for_each_entry(wq, &workqueues, list)
+		wq_update_unbound_numa(wq, cpu, true);
 
-		mutex_unlock(&wq_pool_mutex);
-		break;
-	}
-	return NOTIFY_OK;
+	mutex_unlock(&wq_pool_mutex);
+	return 0;
 }
 
-/*
- * Workqueues should be brought down after normal priority CPU notifiers.
- * This will be registered as low priority CPU notifier.
- */
-static int workqueue_cpu_down_callback(struct notifier_block *nfb,
-						 unsigned long action,
-						 void *hcpu)
+int workqueue_offline_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
 	struct work_struct unbind_work;
 	struct workqueue_struct *wq;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
-		/* unbinding per-cpu workers should happen on the local CPU */
-		INIT_WORK_ONSTACK(&unbind_work, wq_unbind_fn);
-		queue_work_on(cpu, system_highpri_wq, &unbind_work);
-
-		/* update NUMA affinity of unbound workqueues */
-		mutex_lock(&wq_pool_mutex);
-		list_for_each_entry(wq, &workqueues, list)
-			wq_update_unbound_numa(wq, cpu, false);
-		mutex_unlock(&wq_pool_mutex);
-
-		/* wait for per-cpu unbinding to finish */
-		flush_work(&unbind_work);
-		destroy_work_on_stack(&unbind_work);
-		break;
-	}
-	return NOTIFY_OK;
+	/* unbinding per-cpu workers should happen on the local CPU */
+	INIT_WORK_ONSTACK(&unbind_work, wq_unbind_fn);
+	queue_work_on(cpu, system_highpri_wq, &unbind_work);
+
+	/* update NUMA affinity of unbound workqueues */
+	mutex_lock(&wq_pool_mutex);
+	list_for_each_entry(wq, &workqueues, list)
+		wq_update_unbound_numa(wq, cpu, false);
+	mutex_unlock(&wq_pool_mutex);
+
+	/* wait for per-cpu unbinding to finish */
+	flush_work(&unbind_work);
+	destroy_work_on_stack(&unbind_work);
+	return 0;
 }
 
 #ifdef CONFIG_SMP
@@ -5490,9 +5471,6 @@ static int __init init_workqueues(void)
 
 	pwq_cache = KMEM_CACHE(pool_workqueue, SLAB_PANIC);
 
-	cpu_notifier(workqueue_cpu_up_callback, CPU_PRI_WORKQUEUE_UP);
-	hotcpu_notifier(workqueue_cpu_down_callback, CPU_PRI_WORKQUEUE_DOWN);
-
 	wq_numa_init();
 
 	/* initialize CPU pools */

[patch V2 29/67] x86/hpet: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0028-x86-hpet-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 4114 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Jan Beulich <JBeulich@suse.com>
Cc: Jiang Liu <jiang.liu@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/kernel/hpet.c     | 69 +++++++++++++++++++++++-----------------------
 include/linux/cpuhotplug.h |  2 ++
 2 files changed, 36 insertions(+), 35 deletions(-)

diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index f112af7..3d74707 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -710,31 +710,29 @@ static void hpet_work(struct work_struct *w)
 	complete(&hpet_work->complete);
 }
 
-static int hpet_cpuhp_notify(struct notifier_block *n,
-		unsigned long action, void *hcpu)
+static int hpet_cpuhp_online(unsigned int cpu)
 {
-	unsigned long cpu = (unsigned long)hcpu;
 	struct hpet_work_struct work;
+
+	INIT_DELAYED_WORK_ONSTACK(&work.work, hpet_work);
+	init_completion(&work.complete);
+	/* FIXME: add schedule_work_on() */
+	schedule_delayed_work_on(cpu, &work.work, 0);
+	wait_for_completion(&work.complete);
+	destroy_delayed_work_on_stack(&work.work);
+	return 0;
+}
+
+static int hpet_cpuhp_dead(unsigned int cpu)
+{
 	struct hpet_dev *hdev = per_cpu(cpu_hpet_dev, cpu);
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-		INIT_DELAYED_WORK_ONSTACK(&work.work, hpet_work);
-		init_completion(&work.complete);
-		/* FIXME: add schedule_work_on() */
-		schedule_delayed_work_on(cpu, &work.work, 0);
-		wait_for_completion(&work.complete);
-		destroy_delayed_work_on_stack(&work.work);
-		break;
-	case CPU_DEAD:
-		if (hdev) {
-			free_irq(hdev->irq, hdev);
-			hdev->flags &= ~HPET_DEV_USED;
-			per_cpu(cpu_hpet_dev, cpu) = NULL;
-		}
-		break;
-	}
-	return NOTIFY_OK;
+	if (!hdev)
+		return 0;
+	free_irq(hdev->irq, hdev);
+	hdev->flags &= ~HPET_DEV_USED;
+	per_cpu(cpu_hpet_dev, cpu) = NULL;
+	return 0;
 }
 #else
 
@@ -750,11 +748,8 @@ static void hpet_reserve_msi_timers(struct hpet_data *hd)
 }
 #endif
 
-static int hpet_cpuhp_notify(struct notifier_block *n,
-		unsigned long action, void *hcpu)
-{
-	return NOTIFY_OK;
-}
+#define hpet_cpuhp_online	NULL
+#define hpet_cpuhp_dead		NULL
 
 #endif
 
@@ -931,7 +926,7 @@ out_nohpet:
  */
 static __init int hpet_late_init(void)
 {
-	int cpu;
+	int ret;
 
 	if (boot_hpet_disable)
 		return -ENODEV;
@@ -961,16 +956,20 @@ static __init int hpet_late_init(void)
 	if (boot_cpu_has(X86_FEATURE_ARAT))
 		return 0;
 
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu) {
-		hpet_cpuhp_notify(NULL, CPU_ONLINE, (void *)(long)cpu);
-	}
-
 	/* This notifier should be called after workqueue is ready */
-	__hotcpu_notifier(hpet_cpuhp_notify, -20);
-	cpu_notifier_register_done();
-
+	ret = cpuhp_setup_state(CPUHP_AP_X86_HPET_ONLINE, "AP_X86_HPET_ONLINE",
+				hpet_cpuhp_online, NULL);
+	if (ret)
+		return ret;
+	ret = cpuhp_setup_state(CPUHP_X86_HPET_DEAD, "X86_HPET_DEAD", NULL,
+				hpet_cpuhp_dead);
+	if (ret)
+		goto err_cpuhp;
 	return 0;
+
+err_cpuhp:
+	cpuhp_remove_state(CPUHP_AP_X86_HPET_ONLINE);
+	return ret;
 }
 fs_initcall(hpet_late_init);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 60557a9..7449081 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,7 @@ enum cpuhp_state {
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
+	CPUHP_X86_HPET_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
@@ -54,6 +55,7 @@ enum cpuhp_state {
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
+	CPUHP_AP_X86_HPET_ONLINE,
 	CPUHP_AP_ACTIVE,
 	CPUHP_ONLINE,
 };
-- 
2.8.1

[tip:smp/hotplug] x86/hpet: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  48d7f6c715782a6f6d4a5535f1233b764ce6b6f2
Gitweb:     http://git.kernel.org/tip/48d7f6c715782a6f6d4a5535f1233b764ce6b6f2
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:30 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 14 Jul 2016 09:34:44 +0200

x86/hpet: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Jan Beulich <JBeulich@suse.com>
Cc: Jiang Liu <jiang.liu@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.279718463@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/hpet.c     | 69 +++++++++++++++++++++++-----------------------
 include/linux/cpuhotplug.h |  2 ++
 2 files changed, 36 insertions(+), 35 deletions(-)

diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index f112af7..3d74707 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -710,31 +710,29 @@ static void hpet_work(struct work_struct *w)
 	complete(&hpet_work->complete);
 }
 
-static int hpet_cpuhp_notify(struct notifier_block *n,
-		unsigned long action, void *hcpu)
+static int hpet_cpuhp_online(unsigned int cpu)
 {
-	unsigned long cpu = (unsigned long)hcpu;
 	struct hpet_work_struct work;
+
+	INIT_DELAYED_WORK_ONSTACK(&work.work, hpet_work);
+	init_completion(&work.complete);
+	/* FIXME: add schedule_work_on() */
+	schedule_delayed_work_on(cpu, &work.work, 0);
+	wait_for_completion(&work.complete);
+	destroy_delayed_work_on_stack(&work.work);
+	return 0;
+}
+
+static int hpet_cpuhp_dead(unsigned int cpu)
+{
 	struct hpet_dev *hdev = per_cpu(cpu_hpet_dev, cpu);
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-		INIT_DELAYED_WORK_ONSTACK(&work.work, hpet_work);
-		init_completion(&work.complete);
-		/* FIXME: add schedule_work_on() */
-		schedule_delayed_work_on(cpu, &work.work, 0);
-		wait_for_completion(&work.complete);
-		destroy_delayed_work_on_stack(&work.work);
-		break;
-	case CPU_DEAD:
-		if (hdev) {
-			free_irq(hdev->irq, hdev);
-			hdev->flags &= ~HPET_DEV_USED;
-			per_cpu(cpu_hpet_dev, cpu) = NULL;
-		}
-		break;
-	}
-	return NOTIFY_OK;
+	if (!hdev)
+		return 0;
+	free_irq(hdev->irq, hdev);
+	hdev->flags &= ~HPET_DEV_USED;
+	per_cpu(cpu_hpet_dev, cpu) = NULL;
+	return 0;
 }
 #else
 
@@ -750,11 +748,8 @@ static void hpet_reserve_msi_timers(struct hpet_data *hd)
 }
 #endif
 
-static int hpet_cpuhp_notify(struct notifier_block *n,
-		unsigned long action, void *hcpu)
-{
-	return NOTIFY_OK;
-}
+#define hpet_cpuhp_online	NULL
+#define hpet_cpuhp_dead		NULL
 
 #endif
 
@@ -931,7 +926,7 @@ out_nohpet:
  */
 static __init int hpet_late_init(void)
 {
-	int cpu;
+	int ret;
 
 	if (boot_hpet_disable)
 		return -ENODEV;
@@ -961,16 +956,20 @@ static __init int hpet_late_init(void)
 	if (boot_cpu_has(X86_FEATURE_ARAT))
 		return 0;
 
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu) {
-		hpet_cpuhp_notify(NULL, CPU_ONLINE, (void *)(long)cpu);
-	}
-
 	/* This notifier should be called after workqueue is ready */
-	__hotcpu_notifier(hpet_cpuhp_notify, -20);
-	cpu_notifier_register_done();
-
+	ret = cpuhp_setup_state(CPUHP_AP_X86_HPET_ONLINE, "AP_X86_HPET_ONLINE",
+				hpet_cpuhp_online, NULL);
+	if (ret)
+		return ret;
+	ret = cpuhp_setup_state(CPUHP_X86_HPET_DEAD, "X86_HPET_DEAD", NULL,
+				hpet_cpuhp_dead);
+	if (ret)
+		goto err_cpuhp;
 	return 0;
+
+err_cpuhp:
+	cpuhp_remove_state(CPUHP_AP_X86_HPET_ONLINE);
+	return ret;
 }
 fs_initcall(hpet_late_init);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 60557a9..7449081 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -12,6 +12,7 @@ enum cpuhp_state {
 	CPUHP_PERF_BFIN,
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
+	CPUHP_X86_HPET_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
@@ -54,6 +55,7 @@ enum cpuhp_state {
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
+	CPUHP_AP_X86_HPET_ONLINE,
 	CPUHP_AP_ACTIVE,
 	CPUHP_ONLINE,
 };

[patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0029-powerpc-numa-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3687 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/powerpc/mm/numa.c     | 46 ++++++++++++++++------------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 17 insertions(+), 30 deletions(-)

diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index 669a15e..d48ac48 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -581,30 +581,22 @@ static void verify_cpu_node_mapping(int cpu, int node)
 	}
 }
 
-static int cpu_numa_callback(struct notifier_block *nfb, unsigned long action,
-			     void *hcpu)
+/* Must run before sched domains notifier. */
+static int ppc_numa_cpu_prepare(unsigned int cpu)
 {
-	unsigned long lcpu = (unsigned long)hcpu;
-	int ret = NOTIFY_DONE, nid;
+	int nid;
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		nid = numa_setup_cpu(lcpu);
-		verify_cpu_node_mapping((int)lcpu, nid);
-		ret = NOTIFY_OK;
-		break;
+	nid = numa_setup_cpu(cpu);
+	verify_cpu_node_mapping(cpu, nid);
+	return 0;
+}
+
+static int ppc_numa_cpu_dead(unsigned int cpu)
+{
 #ifdef CONFIG_HOTPLUG_CPU
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		unmap_cpu_from_node(lcpu);
-		ret = NOTIFY_OK;
-		break;
+	unmap_cpu_from_node(cpu);
 #endif
-	}
-	return ret;
+	return 0;
 }
 
 /*
@@ -913,11 +905,6 @@ static void __init dump_numa_memory_topology(void)
 	}
 }
 
-static struct notifier_block ppc64_numa_nb = {
-	.notifier_call = cpu_numa_callback,
-	.priority = 1 /* Must run before sched domains notifier. */
-};
-
 /* Initialize NODE_DATA for a node on the local memory */
 static void __init setup_node_data(int nid, u64 start_pfn, u64 end_pfn)
 {
@@ -953,7 +940,7 @@ static void __init setup_node_data(int nid, u64 start_pfn, u64 end_pfn)
 
 void __init initmem_init(void)
 {
-	int nid, cpu;
+	int nid;
 
 	max_low_pfn = memblock_end_of_DRAM() >> PAGE_SHIFT;
 	max_pfn = max_low_pfn;
@@ -985,15 +972,14 @@ void __init initmem_init(void)
 	setup_node_to_cpumask_map();
 
 	reset_numa_cpu_lookup_table();
-	register_cpu_notifier(&ppc64_numa_nb);
+
 	/*
 	 * We need the numa_cpu_lookup_table to be accurate for all CPUs,
 	 * even before we online them, so that we can use cpu_to_{node,mem}
 	 * early in boot, cf. smp_prepare_cpus().
 	 */
-	for_each_present_cpu(cpu) {
-		numa_setup_cpu((unsigned long)cpu);
-	}
+	cpuhp_setup_state(CPUHP_POWER_NUMA_PREPARE, "POWER_NUMA_PREPARE",
+			  ppc_numa_cpu_prepare, ppc_numa_cpu_dead);
 }
 
 static int __init early_numa(char *p)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 7449081..01133ec 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -14,6 +14,7 @@ enum cpuhp_state {
 	CPUHP_PERF_SUPERH,
 	CPUHP_X86_HPET_DEAD,
 	CPUHP_WORKQUEUE_PREP,
+	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
-- 
2.8.1

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Anton Blanchard]

Hi,

> From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> 
> Install the callbacks via the state machine and let the core invoke
> the callbacks on the already online CPUs.

This is causing an oops on ppc64le QEMU, looks like a NULL pointer:

percpu: Embedded 3 pages/cpu @c00000001fe00000 s145816 r0 d50792 u1048576
Unable to handle kernel paging request for data at address 0x00001e08
Faulting instruction address: 0xc0000000001e6b78
Oops: Kernel access of bad area, sig: 11 [#1]
SMP NR_CPUS=2048 NUMA pSeries
Modules linked in:
CPU: 0 PID: 0 Comm: swapper Not tainted 4.7.0-rc7-00198-g425209e #14
task: c000000000d82a00 ti: c000000000dc4000 task.ti: c000000000dc4000
NIP: c0000000001e6b78 LR: c0000000001e6df4 CTR: 0000000000000000
REGS: c000000000dc7b60 TRAP: 0300   Not tainted  (4.7.0-rc7-00198-g425209e)
MSR: 8000000002001033 <SF,VEC,ME,IR,DR,RI,LE>  CR: 44000220  XER: 00000000
CFAR: 0000000000008468 DAR: 0000000000001e08 DSISR: 40000000 SOFTE: 0 
GPR00: 0000000000000006 c000000000dc7de0 c000000000dc6c00 0000000000000000 
GPR04: 0000000000000000 0000000000000000 c00000001fe1fb70 0000000000000010 
GPR08: c000000000dfe918 c000000000e50dd0 c000000000e56c00 0000000000000001 
GPR12: 0000000000000000 c00000000fe00000 0000000000000060 0000000000f1d618 
GPR16: 0000000000efccd8 0000000000efcb20 fffffffffffffffd 0000000000000000 
GPR20: 000000001f150000 c000000000dfa8e0 c000000000ccfafc c000000000dfeb18 
GPR24: c000000000dfee34 c000000000ccfaf8 0000000000000000 0000000000000001 
GPR28: c000000000ebad20 c000000000ccfb00 0000000000000000 c00000001fe1fb00 
NIP [c0000000001e6b78] local_memory_node+0x18/0x80
LR [c0000000001e6df4] __build_all_zonelists+0x214/0x2d0
Call Trace:
[c000000000dc7de0] [c0000000001e6ccc] __build_all_zonelists+0xec/0x2d0 (unreliable)
[c000000000dc7e70] [c000000000c39fbc] build_all_zonelists_init+0x1c/0x3c
[c000000000dc7e90] [c000000000282acc] build_all_zonelists+0x17c/0x18c
[c000000000dc7f00] [c000000000c13c54] start_kernel+0x18c/0x514
[c000000000dc7f90] [c000000000008c60] start_here_common+0x20/0xa0
Instruction dump:
38810178 7f63db78 48769171 60000000 4bfffd2c 60420000 3c4c00be 384200a0 
3d420009 78631f24 392aa1d0 7c69182a <81231e08> 38631e00 2b890002 419d001c 

Anton

> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
> Cc: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Michael Ellerman <mpe@ellerman.id.au>
> Cc: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: linuxppc-dev@lists.ozlabs.org
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
> ---
>  arch/powerpc/mm/numa.c     | 46
> ++++++++++++++++------------------------------
> include/linux/cpuhotplug.h |  1 + 2 files changed, 17 insertions(+),
> 30 deletions(-)
> 
> diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
> index 669a15e..d48ac48 100644
> --- a/arch/powerpc/mm/numa.c
> +++ b/arch/powerpc/mm/numa.c
> @@ -581,30 +581,22 @@ static void verify_cpu_node_mapping(int cpu,
> int node) }
>  }
>  
> -static int cpu_numa_callback(struct notifier_block *nfb, unsigned
> long action,
> -			     void *hcpu)
> +/* Must run before sched domains notifier. */
> +static int ppc_numa_cpu_prepare(unsigned int cpu)
>  {
> -	unsigned long lcpu = (unsigned long)hcpu;
> -	int ret = NOTIFY_DONE, nid;
> +	int nid;
>  
> -	switch (action) {
> -	case CPU_UP_PREPARE:
> -	case CPU_UP_PREPARE_FROZEN:
> -		nid = numa_setup_cpu(lcpu);
> -		verify_cpu_node_mapping((int)lcpu, nid);
> -		ret = NOTIFY_OK;
> -		break;
> +	nid = numa_setup_cpu(cpu);
> +	verify_cpu_node_mapping(cpu, nid);
> +	return 0;
> +}
> +
> +static int ppc_numa_cpu_dead(unsigned int cpu)
> +{
>  #ifdef CONFIG_HOTPLUG_CPU
> -	case CPU_DEAD:
> -	case CPU_DEAD_FROZEN:
> -	case CPU_UP_CANCELED:
> -	case CPU_UP_CANCELED_FROZEN:
> -		unmap_cpu_from_node(lcpu);
> -		ret = NOTIFY_OK;
> -		break;
> +	unmap_cpu_from_node(cpu);
>  #endif
> -	}
> -	return ret;
> +	return 0;
>  }
>  
>  /*
> @@ -913,11 +905,6 @@ static void __init
> dump_numa_memory_topology(void) }
>  }
>  
> -static struct notifier_block ppc64_numa_nb = {
> -	.notifier_call = cpu_numa_callback,
> -	.priority = 1 /* Must run before sched domains notifier. */
> -};
> -
>  /* Initialize NODE_DATA for a node on the local memory */
>  static void __init setup_node_data(int nid, u64 start_pfn, u64
> end_pfn) {
> @@ -953,7 +940,7 @@ static void __init setup_node_data(int nid, u64
> start_pfn, u64 end_pfn) 
>  void __init initmem_init(void)
>  {
> -	int nid, cpu;
> +	int nid;
>  
>  	max_low_pfn = memblock_end_of_DRAM() >> PAGE_SHIFT;
>  	max_pfn = max_low_pfn;
> @@ -985,15 +972,14 @@ void __init initmem_init(void)
>  	setup_node_to_cpumask_map();
>  
>  	reset_numa_cpu_lookup_table();
> -	register_cpu_notifier(&ppc64_numa_nb);
> +
>  	/*
>  	 * We need the numa_cpu_lookup_table to be accurate for all
> CPUs,
>  	 * even before we online them, so that we can use
> cpu_to_{node,mem}
>  	 * early in boot, cf. smp_prepare_cpus().
>  	 */
> -	for_each_present_cpu(cpu) {
> -		numa_setup_cpu((unsigned long)cpu);
> -	}
> +	cpuhp_setup_state(CPUHP_POWER_NUMA_PREPARE,
> "POWER_NUMA_PREPARE",
> +			  ppc_numa_cpu_prepare, ppc_numa_cpu_dead);
>  }
>  
>  static int __init early_numa(char *p)
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index 7449081..01133ec 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -14,6 +14,7 @@ enum cpuhp_state {
>  	CPUHP_PERF_SUPERH,
>  	CPUHP_X86_HPET_DEAD,
>  	CPUHP_WORKQUEUE_PREP,
> +	CPUHP_POWER_NUMA_PREPARE,
>  	CPUHP_NOTIFY_PREPARE,
>  	CPUHP_BRINGUP_CPU,
>  	CPUHP_AP_IDLE_DEAD,

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Anna-Maria Gleixner]

Hi,

On 2016-07-14 23:42, Anton Blanchard wrote:
> Hi,
>
>> From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
>>
>> Install the callbacks via the state machine and let the core invoke
>> the callbacks on the already online CPUs.
>
> This is causing an oops on ppc64le QEMU, looks like a NULL pointer:

Did you tested it against tip WIP.hotplug?

Regards,
      Anna-Maria

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Anton Blanchard]

Hi Anna-Maria,

> >> Install the callbacks via the state machine and let the core invoke
> >> the callbacks on the already online CPUs.  
> >
> > This is causing an oops on ppc64le QEMU, looks like a NULL
> > pointer:  
> 
> Did you tested it against tip WIP.hotplug?

I noticed tip started failing in my CI environment which tests on QEMU.
The failure bisected to commit 425209e0abaf2c6e3a90ce4fedb935c10652bf80

It reproduces running ppc64le QEMU on a x86-64 box. On Ubuntu:

sudo apt-get install qemu-system-ppc gcc-powerpc64le-linux-gnu

make ARCH=powerpc pseries_le_defconfig

make ARCH=powerpc CROSS_COMPILE=powerpc64le-linux-gnu- vmlinux -j4

qemu-system-ppc64 -nographic -vga none -kernel vmlinux 

Anton

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Ingo Molnar]

* Anton Blanchard <anton@samba.org> wrote:

> Hi Anna-Maria,
> 
> > >> Install the callbacks via the state machine and let the core invoke
> > >> the callbacks on the already online CPUs.  
> > >
> > > This is causing an oops on ppc64le QEMU, looks like a NULL
> > > pointer:  
> > 
> > Did you tested it against tip WIP.hotplug?
> 
> I noticed tip started failing in my CI environment which tests on QEMU.
> The failure bisected to commit 425209e0abaf2c6e3a90ce4fedb935c10652bf80

That's very useful, thanks Anton!

I have removed this commit from the series for the time being, refactored the 
followup commits (there was one trivial conflict). We can re-try this patch when a 
fix is found.

Thanks,

	Ingo

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Anton Blanchard]

> > I noticed tip started failing in my CI environment which tests on
> > QEMU. The failure bisected to commit
> > 425209e0abaf2c6e3a90ce4fedb935c10652bf80  
> 
> That's very useful, thanks Anton!
> 
> I have removed this commit from the series for the time being,
> refactored the followup commits (there was one trivial conflict). We
> can re-try this patch when a fix is found.

Thanks Ingo, my tests are passing again after your last push.

Anton

Re: [patch V2 30/67] powerpc/numa: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

* Anton Blanchard | 2016-07-15 10:28:25 [+1000]:

>Hi Anna-Maria,
Hi Anton,

>> >> Install the callbacks via the state machine and let the core invoke
>> >> the callbacks on the already online CPUs.  
>> >
>> > This is causing an oops on ppc64le QEMU, looks like a NULL
>> > pointer:  
>> 
>> Did you tested it against tip WIP.hotplug?
>
>I noticed tip started failing in my CI environment which tests on QEMU.
>The failure bisected to commit 425209e0abaf2c6e3a90ce4fedb935c10652bf80
>
>It reproduces running ppc64le QEMU on a x86-64 box. On Ubuntu:
…

Thanks for that. I can reproduce this ontop of latest WIP.hotplug with
this patch.

>Anton

Sebastian

[PATCH v2] powerpc/numa: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

v1…v2: manual callback invocation on boot-CPU (cpuhp is not up yet and
       we need them all before additional CPUs are up).

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/powerpc/mm/numa.c     | 48 ++++++++++++++++++----------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 20 insertions(+), 29 deletions(-)

diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index 669a15e7fa76..6dc07ddbfd04 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -581,30 +581,22 @@ static void verify_cpu_node_mapping(int cpu, int node)
 	}
 }
 
-static int cpu_numa_callback(struct notifier_block *nfb, unsigned long action,
-			     void *hcpu)
+/* Must run before sched domains notifier. */
+static int ppc_numa_cpu_prepare(unsigned int cpu)
 {
-	unsigned long lcpu = (unsigned long)hcpu;
-	int ret = NOTIFY_DONE, nid;
+	int nid;
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		nid = numa_setup_cpu(lcpu);
-		verify_cpu_node_mapping((int)lcpu, nid);
-		ret = NOTIFY_OK;
-		break;
+	nid = numa_setup_cpu(cpu);
+	verify_cpu_node_mapping(cpu, nid);
+	return 0;
+}
+
+static int ppc_numa_cpu_dead(unsigned int cpu)
+{
 #ifdef CONFIG_HOTPLUG_CPU
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		unmap_cpu_from_node(lcpu);
-		ret = NOTIFY_OK;
-		break;
+	unmap_cpu_from_node(cpu);
 #endif
-	}
-	return ret;
+	return 0;
 }
 
 /*
@@ -913,11 +905,6 @@ static void __init dump_numa_memory_topology(void)
 	}
 }
 
-static struct notifier_block ppc64_numa_nb = {
-	.notifier_call = cpu_numa_callback,
-	.priority = 1 /* Must run before sched domains notifier. */
-};
-
 /* Initialize NODE_DATA for a node on the local memory */
 static void __init setup_node_data(int nid, u64 start_pfn, u64 end_pfn)
 {
@@ -985,15 +972,18 @@ void __init initmem_init(void)
 	setup_node_to_cpumask_map();
 
 	reset_numa_cpu_lookup_table();
-	register_cpu_notifier(&ppc64_numa_nb);
+
 	/*
 	 * We need the numa_cpu_lookup_table to be accurate for all CPUs,
 	 * even before we online them, so that we can use cpu_to_{node,mem}
 	 * early in boot, cf. smp_prepare_cpus().
+	 * _nocalls() + manual invocation is used because cpuhp is not yet
+	 * initialized for the boot CPU.
 	 */
-	for_each_present_cpu(cpu) {
-		numa_setup_cpu((unsigned long)cpu);
-	}
+	cpuhp_setup_state_nocalls(CPUHP_POWER_NUMA_PREPARE, "POWER_NUMA_PREPARE",
+				  ppc_numa_cpu_prepare, ppc_numa_cpu_dead);
+	for_each_present_cpu(cpu)
+		numa_setup_cpu(cpu);
 }
 
 static int __init early_numa(char *p)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 09ef54bcba39..5015f463d314 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -15,6 +15,7 @@ enum cpuhp_state {
 	CPUHP_X86_HPET_DEAD,
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
+	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
-- 
2.8.1

[tip:smp/hotplug] powerpc/numa: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  bdab88e006504cd83fac98705814485cbe3ef5b4
Gitweb:     http://git.kernel.org/tip/bdab88e006504cd83fac98705814485cbe3ef5b4
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Mon, 18 Jul 2016 16:07:28 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 22 Jul 2016 21:53:17 +0200

powerpc/numa: Convert to hotplug state machine

Install the callbacks via the state machine. On the boot cpu the callback is
invoked manually because cpuhp is not up yet and everything must be
preinitialized before additional CPUs are up.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Cc: Anton Blanchard <anton@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160718140727.GA13132@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 arch/powerpc/mm/numa.c     | 48 ++++++++++++++++++----------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 20 insertions(+), 29 deletions(-)

diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index 669a15e..6dc07dd 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -581,30 +581,22 @@ static void verify_cpu_node_mapping(int cpu, int node)
 	}
 }
 
-static int cpu_numa_callback(struct notifier_block *nfb, unsigned long action,
-			     void *hcpu)
+/* Must run before sched domains notifier. */
+static int ppc_numa_cpu_prepare(unsigned int cpu)
 {
-	unsigned long lcpu = (unsigned long)hcpu;
-	int ret = NOTIFY_DONE, nid;
+	int nid;
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		nid = numa_setup_cpu(lcpu);
-		verify_cpu_node_mapping((int)lcpu, nid);
-		ret = NOTIFY_OK;
-		break;
+	nid = numa_setup_cpu(cpu);
+	verify_cpu_node_mapping(cpu, nid);
+	return 0;
+}
+
+static int ppc_numa_cpu_dead(unsigned int cpu)
+{
 #ifdef CONFIG_HOTPLUG_CPU
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		unmap_cpu_from_node(lcpu);
-		ret = NOTIFY_OK;
-		break;
+	unmap_cpu_from_node(cpu);
 #endif
-	}
-	return ret;
+	return 0;
 }
 
 /*
@@ -913,11 +905,6 @@ static void __init dump_numa_memory_topology(void)
 	}
 }
 
-static struct notifier_block ppc64_numa_nb = {
-	.notifier_call = cpu_numa_callback,
-	.priority = 1 /* Must run before sched domains notifier. */
-};
-
 /* Initialize NODE_DATA for a node on the local memory */
 static void __init setup_node_data(int nid, u64 start_pfn, u64 end_pfn)
 {
@@ -985,15 +972,18 @@ void __init initmem_init(void)
 	setup_node_to_cpumask_map();
 
 	reset_numa_cpu_lookup_table();
-	register_cpu_notifier(&ppc64_numa_nb);
+
 	/*
 	 * We need the numa_cpu_lookup_table to be accurate for all CPUs,
 	 * even before we online them, so that we can use cpu_to_{node,mem}
 	 * early in boot, cf. smp_prepare_cpus().
+	 * _nocalls() + manual invocation is used because cpuhp is not yet
+	 * initialized for the boot CPU.
 	 */
-	for_each_present_cpu(cpu) {
-		numa_setup_cpu((unsigned long)cpu);
-	}
+	cpuhp_setup_state_nocalls(CPUHP_POWER_NUMA_PREPARE, "POWER_NUMA_PREPARE",
+				  ppc_numa_cpu_prepare, ppc_numa_cpu_dead);
+	for_each_present_cpu(cpu)
+		numa_setup_cpu(cpu);
 }
 
 static int __init early_numa(char *p)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 09ef54b..5015f46 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -15,6 +15,7 @@ enum cpuhp_state {
 	CPUHP_X86_HPET_DEAD,
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
+	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,

[patch V2 31/67] KVM/x86: Remove superfluous SMP function call

[Anna-Maria Gleixner]

[-- Attachment #1: 0030-KVM-x86-Remove-superfluous-SMP-function-call.patch --]
[-- Type: text/plain, Size: 1572 bytes --]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

Since the following commit:

  1cf4f629d9d2 ("cpu/hotplug: Move online calls to hotplugged cpu")

... the CPU_ONLINE and CPU_DOWN_PREPARE notifiers are always run on the hot
plugged CPU, and as of commit:

  3b9d6da67e11 ("cpu/hotplug: Fix rollback during error-out in __cpu_disable()")

the CPU_DOWN_FAILED notifier also runs on the hot plugged CPU.  This patch
converts the SMP functional calls into direct calls.

smp_function_call_single() executes the function with interrupts
disabled. This calling convention is not preserved because there
is no reason to do so.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
---
 arch/x86/kvm/x86.c |    6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5662,15 +5662,13 @@ static struct notifier_block kvmclock_cp
 static int kvmclock_cpu_notifier(struct notifier_block *nfb,
 					unsigned long action, void *hcpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	switch (action) {
 		case CPU_ONLINE:
 		case CPU_DOWN_FAILED:
-			smp_call_function_single(cpu, tsc_khz_changed, NULL, 1);
+			tsc_khz_changed(NULL);
 			break;
 		case CPU_DOWN_PREPARE:
-			smp_call_function_single(cpu, tsc_bad, NULL, 1);
+			tsc_bad(NULL);
 			break;
 	}
 	return NOTIFY_OK;

Re: [patch V2 31/67] KVM/x86: Remove superfluous SMP function call

[Paolo Bonzini]

On 13/07/2016 19:16, Anna-Maria Gleixner wrote:
> From: Anna-Maria Gleixner <anna-maria@linutronix.de>
> 
> Since the following commit:
> 
>   1cf4f629d9d2 ("cpu/hotplug: Move online calls to hotplugged cpu")
> 
> ... the CPU_ONLINE and CPU_DOWN_PREPARE notifiers are always run on the hot
> plugged CPU, and as of commit:
> 
>   3b9d6da67e11 ("cpu/hotplug: Fix rollback during error-out in __cpu_disable()")
> 
> the CPU_DOWN_FAILED notifier also runs on the hot plugged CPU.  This patch
> converts the SMP functional calls into direct calls.
> 
> smp_function_call_single() executes the function with interrupts
> disabled. This calling convention is not preserved because there
> is no reason to do so.
> 
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Radim Krcmar <rkrcmar@redhat.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: kvm@vger.kernel.org
> ---
>  arch/x86/kvm/x86.c |    6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -5662,15 +5662,13 @@ static struct notifier_block kvmclock_cp
>  static int kvmclock_cpu_notifier(struct notifier_block *nfb,
>  					unsigned long action, void *hcpu)
>  {
> -	unsigned int cpu = (unsigned long)hcpu;
> -
>  	switch (action) {
>  		case CPU_ONLINE:
>  		case CPU_DOWN_FAILED:
> -			smp_call_function_single(cpu, tsc_khz_changed, NULL, 1);
> +			tsc_khz_changed(NULL);
>  			break;
>  		case CPU_DOWN_PREPARE:
> -			smp_call_function_single(cpu, tsc_bad, NULL, 1);
> +			tsc_bad(NULL);
>  			break;
>  	}
>  	return NOTIFY_OK;
> 
> 

Acked-by: Paolo Bonzini <pbonzini@redhat.com>

[tip:smp/hotplug] KVM/x86: Remove superfluous SMP function call

[tip-bot for Anna-Maria Gleixner]

Commit-ID:  162e52a117af5d9c5dfc62b7c2ee90b3e91a7b32
Gitweb:     http://git.kernel.org/tip/162e52a117af5d9c5dfc62b7c2ee90b3e91a7b32
Author:     Anna-Maria Gleixner <anna-maria@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:32 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:21 +0200

KVM/x86: Remove superfluous SMP function call

Since the following commit:

  1cf4f629d9d2 ("cpu/hotplug: Move online calls to hotplugged cpu")

... the CPU_ONLINE and CPU_DOWN_PREPARE notifiers are always run on the hot
plugged CPU, and as of commit:

  3b9d6da67e11 ("cpu/hotplug: Fix rollback during error-out in __cpu_disable()")

the CPU_DOWN_FAILED notifier also runs on the hot plugged CPU.  This patch
converts the SMP functional calls into direct calls.

smp_function_call_single() executes the function with interrupts
disabled. This calling convention is not preserved because there
is no reason to do so.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.452527104@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kvm/x86.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 7da5dd2..6f90595 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5662,15 +5662,13 @@ static struct notifier_block kvmclock_cpufreq_notifier_block = {
 static int kvmclock_cpu_notifier(struct notifier_block *nfb,
 					unsigned long action, void *hcpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	switch (action) {
 		case CPU_ONLINE:
 		case CPU_DOWN_FAILED:
-			smp_call_function_single(cpu, tsc_khz_changed, NULL, 1);
+			tsc_khz_changed(NULL);
 			break;
 		case CPU_DOWN_PREPARE:
-			smp_call_function_single(cpu, tsc_bad, NULL, 1);
+			tsc_bad(NULL);
 			break;
 	}
 	return NOTIFY_OK;

[patch V2 32/67] x86/kvm/kvmclock: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0031-x86-kvm-kvmclock-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3391 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

We assumed that the priority ordering was ment to invoke the online
callback as the last step. In the original code this also invoked the
down prepare callback as the last step. With the symmetric state
machine the down prepare callback is now the first step.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/kvm/x86.c         |   33 ++++++++-------------------------
 include/linux/cpuhotplug.h |    1 +
 2 files changed, 9 insertions(+), 25 deletions(-)

--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5552,9 +5552,10 @@ int kvm_fast_pio_out(struct kvm_vcpu *vc
 }
 EXPORT_SYMBOL_GPL(kvm_fast_pio_out);
 
-static void tsc_bad(void *info)
+static int kvmclock_cpu_down_prep(unsigned int cpu)
 {
 	__this_cpu_write(cpu_tsc_khz, 0);
+	return 0;
 }
 
 static void tsc_khz_changed(void *data)
@@ -5659,33 +5660,18 @@ static struct notifier_block kvmclock_cp
 	.notifier_call  = kvmclock_cpufreq_notifier
 };
 
-static int kvmclock_cpu_notifier(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int kvmclock_cpu_online(unsigned int cpu)
 {
-	switch (action) {
-		case CPU_ONLINE:
-		case CPU_DOWN_FAILED:
-			tsc_khz_changed(NULL);
-			break;
-		case CPU_DOWN_PREPARE:
-			tsc_bad(NULL);
-			break;
-	}
-	return NOTIFY_OK;
+	tsc_khz_changed(NULL);
+	return 0;
 }
 
-static struct notifier_block kvmclock_cpu_notifier_block = {
-	.notifier_call  = kvmclock_cpu_notifier,
-	.priority = -INT_MAX
-};
-
 static void kvm_timer_init(void)
 {
 	int cpu;
 
 	max_tsc_khz = tsc_khz;
 
-	cpu_notifier_register_begin();
 	if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) {
 #ifdef CONFIG_CPU_FREQ
 		struct cpufreq_policy policy;
@@ -5700,12 +5686,9 @@ static void kvm_timer_init(void)
 					  CPUFREQ_TRANSITION_NOTIFIER);
 	}
 	pr_debug("kvm: max_tsc_khz = %ld\n", max_tsc_khz);
-	for_each_online_cpu(cpu)
-		smp_call_function_single(cpu, tsc_khz_changed, NULL, 1);
-
-	__register_hotcpu_notifier(&kvmclock_cpu_notifier_block);
-	cpu_notifier_register_done();
 
+	cpuhp_setup_state(CPUHP_AP_X86_KVM_CLK_ONLINE, "AP_X86_KVM_CLK_ONLINE",
+			  kvmclock_cpu_online, kvmclock_cpu_down_prep);
 }
 
 static DEFINE_PER_CPU(struct kvm_vcpu *, current_vcpu);
@@ -5894,7 +5877,7 @@ void kvm_arch_exit(void)
 	if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC))
 		cpufreq_unregister_notifier(&kvmclock_cpufreq_notifier_block,
 					    CPUFREQ_TRANSITION_NOTIFIER);
-	unregister_hotcpu_notifier(&kvmclock_cpu_notifier_block);
+	cpuhp_remove_state_nocalls(CPUHP_AP_X86_KVM_CLK_ONLINE);
 #ifdef CONFIG_X86_64
 	pvclock_gtod_unregister_notifier(&pvclock_gtod_notifier);
 #endif
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -57,6 +57,7 @@ enum cpuhp_state {
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
 	CPUHP_AP_X86_HPET_ONLINE,
+	CPUHP_AP_X86_KVM_CLK_ONLINE,
 	CPUHP_AP_ACTIVE,
 	CPUHP_ONLINE,
 };

[tip:smp/hotplug] x86/kvm/kvmclock: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  251a5fd64b720a0646eddb8c54f5e9ddba066e72
Gitweb:     http://git.kernel.org/tip/251a5fd64b720a0646eddb8c54f5e9ddba066e72
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:33 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:21 +0200

x86/kvm/kvmclock: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

We assumed that the priority ordering was ment to invoke the online
callback as the last step. In the original code this also invoked the
down prepare callback as the last step. With the symmetric state
machine the down prepare callback is now the first step.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.542880859@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kvm/x86.c         | 33 ++++++++-------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 9 insertions(+), 25 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 6f90595..f899127 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5552,9 +5552,10 @@ int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port)
 }
 EXPORT_SYMBOL_GPL(kvm_fast_pio_out);
 
-static void tsc_bad(void *info)
+static int kvmclock_cpu_down_prep(unsigned int cpu)
 {
 	__this_cpu_write(cpu_tsc_khz, 0);
+	return 0;
 }
 
 static void tsc_khz_changed(void *data)
@@ -5659,33 +5660,18 @@ static struct notifier_block kvmclock_cpufreq_notifier_block = {
 	.notifier_call  = kvmclock_cpufreq_notifier
 };
 
-static int kvmclock_cpu_notifier(struct notifier_block *nfb,
-					unsigned long action, void *hcpu)
+static int kvmclock_cpu_online(unsigned int cpu)
 {
-	switch (action) {
-		case CPU_ONLINE:
-		case CPU_DOWN_FAILED:
-			tsc_khz_changed(NULL);
-			break;
-		case CPU_DOWN_PREPARE:
-			tsc_bad(NULL);
-			break;
-	}
-	return NOTIFY_OK;
+	tsc_khz_changed(NULL);
+	return 0;
 }
 
-static struct notifier_block kvmclock_cpu_notifier_block = {
-	.notifier_call  = kvmclock_cpu_notifier,
-	.priority = -INT_MAX
-};
-
 static void kvm_timer_init(void)
 {
 	int cpu;
 
 	max_tsc_khz = tsc_khz;
 
-	cpu_notifier_register_begin();
 	if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) {
 #ifdef CONFIG_CPU_FREQ
 		struct cpufreq_policy policy;
@@ -5700,12 +5686,9 @@ static void kvm_timer_init(void)
 					  CPUFREQ_TRANSITION_NOTIFIER);
 	}
 	pr_debug("kvm: max_tsc_khz = %ld\n", max_tsc_khz);
-	for_each_online_cpu(cpu)
-		smp_call_function_single(cpu, tsc_khz_changed, NULL, 1);
-
-	__register_hotcpu_notifier(&kvmclock_cpu_notifier_block);
-	cpu_notifier_register_done();
 
+	cpuhp_setup_state(CPUHP_AP_X86_KVM_CLK_ONLINE, "AP_X86_KVM_CLK_ONLINE",
+			  kvmclock_cpu_online, kvmclock_cpu_down_prep);
 }
 
 static DEFINE_PER_CPU(struct kvm_vcpu *, current_vcpu);
@@ -5894,7 +5877,7 @@ void kvm_arch_exit(void)
 	if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC))
 		cpufreq_unregister_notifier(&kvmclock_cpufreq_notifier_block,
 					    CPUFREQ_TRANSITION_NOTIFIER);
-	unregister_hotcpu_notifier(&kvmclock_cpu_notifier_block);
+	cpuhp_remove_state_nocalls(CPUHP_AP_X86_KVM_CLK_ONLINE);
 #ifdef CONFIG_X86_64
 	pvclock_gtod_unregister_notifier(&pvclock_gtod_notifier);
 #endif
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 7449081..8c190a8 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -56,6 +56,7 @@ enum cpuhp_state {
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
 	CPUHP_AP_X86_HPET_ONLINE,
+	CPUHP_AP_X86_KVM_CLK_ONLINE,
 	CPUHP_AP_ACTIVE,
 	CPUHP_ONLINE,
 };

[patch V2 33/67] x86/apb_timer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0032-x86-apb_timer-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2720 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine. There is no setup just one
teardown callback. Remove the silly comment about the workqueue up dependency.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/kernel/apb_timer.c | 29 ++++++++++-------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 11 insertions(+), 19 deletions(-)

diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
index cefacba..456316f 100644
--- a/arch/x86/kernel/apb_timer.c
+++ b/arch/x86/kernel/apb_timer.c
@@ -215,26 +215,18 @@ void apbt_setup_secondary_clock(void)
  * cpu timers during the offline process due to the ordering of notification.
  * the extra interrupt is harmless.
  */
-static int apbt_cpuhp_notify(struct notifier_block *n,
-			     unsigned long action, void *hcpu)
+static int apbt_cpu_dead(unsigned int cpu)
 {
-	unsigned long cpu = (unsigned long)hcpu;
 	struct apbt_dev *adev = &per_cpu(cpu_apbt_dev, cpu);
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DEAD:
-		dw_apb_clockevent_pause(adev->timer);
-		if (system_state == SYSTEM_RUNNING) {
-			pr_debug("skipping APBT CPU %lu offline\n", cpu);
-		} else {
-			pr_debug("APBT clockevent for cpu %lu offline\n", cpu);
-			dw_apb_clockevent_stop(adev->timer);
-		}
-		break;
-	default:
-		pr_debug("APBT notified %lu, no action\n", action);
+	dw_apb_clockevent_pause(adev->timer);
+	if (system_state == SYSTEM_RUNNING) {
+		pr_debug("skipping APBT CPU %u offline\n", cpu);
+	} else {
+		pr_debug("APBT clockevent for cpu %u offline\n", cpu);
+		dw_apb_clockevent_stop(adev->timer);
 	}
-	return NOTIFY_OK;
+	return 0;
 }
 
 static __init int apbt_late_init(void)
@@ -242,9 +234,8 @@ static __init int apbt_late_init(void)
 	if (intel_mid_timer_options == INTEL_MID_TIMER_LAPIC_APBT ||
 		!apb_timer_block_enabled)
 		return 0;
-	/* This notifier should be called after workqueue is ready */
-	hotcpu_notifier(apbt_cpuhp_notify, -20);
-	return 0;
+	return cpuhp_setup_state(CPUHP_X86_APB_DEAD, "X86_APB_DEAD", NULL,
+				 apbt_cpu_dead);
 }
 fs_initcall(apbt_late_init);
 #else
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 7ac99ac..e8d1b84 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -13,6 +13,7 @@ enum cpuhp_state {
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
 	CPUHP_X86_HPET_DEAD,
+	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
-- 
2.8.1

[tip:smp/hotplug] x86/apb_timer: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  148b9e2abea6f50e4620e4ac0683b4913c4364c0
Gitweb:     http://git.kernel.org/tip/148b9e2abea6f50e4620e4ac0683b4913c4364c0
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:34 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:22 +0200

x86/apb_timer: Convert to hotplug state machine

Install the callbacks via the state machine. There is no setup just one
teardown callback. Remove the silly comment about the workqueue up dependency.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.625342983@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apb_timer.c | 29 ++++++++++-------------------
 include/linux/cpuhotplug.h  |  1 +
 2 files changed, 11 insertions(+), 19 deletions(-)

diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
index cefacba..456316f 100644
--- a/arch/x86/kernel/apb_timer.c
+++ b/arch/x86/kernel/apb_timer.c
@@ -215,26 +215,18 @@ void apbt_setup_secondary_clock(void)
  * cpu timers during the offline process due to the ordering of notification.
  * the extra interrupt is harmless.
  */
-static int apbt_cpuhp_notify(struct notifier_block *n,
-			     unsigned long action, void *hcpu)
+static int apbt_cpu_dead(unsigned int cpu)
 {
-	unsigned long cpu = (unsigned long)hcpu;
 	struct apbt_dev *adev = &per_cpu(cpu_apbt_dev, cpu);
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DEAD:
-		dw_apb_clockevent_pause(adev->timer);
-		if (system_state == SYSTEM_RUNNING) {
-			pr_debug("skipping APBT CPU %lu offline\n", cpu);
-		} else {
-			pr_debug("APBT clockevent for cpu %lu offline\n", cpu);
-			dw_apb_clockevent_stop(adev->timer);
-		}
-		break;
-	default:
-		pr_debug("APBT notified %lu, no action\n", action);
+	dw_apb_clockevent_pause(adev->timer);
+	if (system_state == SYSTEM_RUNNING) {
+		pr_debug("skipping APBT CPU %u offline\n", cpu);
+	} else {
+		pr_debug("APBT clockevent for cpu %u offline\n", cpu);
+		dw_apb_clockevent_stop(adev->timer);
 	}
-	return NOTIFY_OK;
+	return 0;
 }
 
 static __init int apbt_late_init(void)
@@ -242,9 +234,8 @@ static __init int apbt_late_init(void)
 	if (intel_mid_timer_options == INTEL_MID_TIMER_LAPIC_APBT ||
 		!apb_timer_block_enabled)
 		return 0;
-	/* This notifier should be called after workqueue is ready */
-	hotcpu_notifier(apbt_cpuhp_notify, -20);
-	return 0;
+	return cpuhp_setup_state(CPUHP_X86_APB_DEAD, "X86_APB_DEAD", NULL,
+				 apbt_cpu_dead);
 }
 fs_initcall(apbt_late_init);
 #else
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 8c190a8..3e719ba 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -13,6 +13,7 @@ enum cpuhp_state {
 	CPUHP_PERF_POWER,
 	CPUHP_PERF_SUPERH,
 	CPUHP_X86_HPET_DEAD,
+	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,

[patch V2 34/67] arm: Convert VFP hotplug notifiers to state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0033-arm-Convert-VFP-hotplug-notifiers-to-state-machine.patch --]
[-- Type: text/plain, Size: 2911 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Straight forward conversion plus commentary why code which is executed
in hotplug callbacks needs to be invoked before installing them.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm/vfp/vfpmodule.c   | 28 +++++++++++++++++-----------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/arch/arm/vfp/vfpmodule.c b/arch/arm/vfp/vfpmodule.c
index 73085d3..da0b33d 100644
--- a/arch/arm/vfp/vfpmodule.c
+++ b/arch/arm/vfp/vfpmodule.c
@@ -643,19 +643,19 @@ int vfp_restore_user_hwstate(struct user_vfp __user *ufp,
  * hardware state at every thread switch.  We clear our held state when
  * a CPU has been killed, indicating that the VFP hardware doesn't contain
  * a threads VFP state.  When a CPU starts up, we re-enable access to the
- * VFP hardware.
- *
- * Both CPU_DYING and CPU_STARTING are called on the CPU which
+ * VFP hardware. The callbacks below are called on the CPU which
  * is being offlined/onlined.
  */
-static int vfp_hotplug(struct notifier_block *b, unsigned long action,
-	void *hcpu)
+static int vfp_dying_cpu(unsigned int cpu)
 {
-	if (action == CPU_DYING || action == CPU_DYING_FROZEN)
-		vfp_current_hw_state[(long)hcpu] = NULL;
-	else if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		vfp_enable(NULL);
-	return NOTIFY_OK;
+	vfp_force_reload(cpu, current_thread_info());
+	return 0;
+}
+
+static int vfp_starting_cpu(unsigned int unused)
+{
+	vfp_enable(NULL);
+	return 0;
 }
 
 void vfp_kmode_exception(void)
@@ -732,6 +732,10 @@ static int __init vfp_init(void)
 	unsigned int vfpsid;
 	unsigned int cpu_arch = cpu_architecture();
 
+	/*
+	 * Enable the access to the VFP on all online CPUs so the
+	 * following test on FPSID will succeed.
+	 */
 	if (cpu_arch >= CPU_ARCH_ARMv6)
 		on_each_cpu(vfp_enable, NULL, 1);
 
@@ -794,7 +798,9 @@ static int __init vfp_init(void)
 		VFP_arch = (vfpsid & FPSID_ARCH_MASK) >> FPSID_ARCH_BIT;
 	}
 
-	hotcpu_notifier(vfp_hotplug, 0);
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_VFP_STARTING,
+				  "AP_ARM_VFP_STARTING", vfp_starting_cpu,
+				  vfp_dying_cpu);
 
 	vfp_vector = vfp_support_entry;
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index e8d1b84..2e227b6 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
+	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] arm: Convert VFP hotplug notifiers to state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  e5b61bafe70477e05e1dce0d6ca4ec181e23cb2a
Gitweb:     http://git.kernel.org/tip/e5b61bafe70477e05e1dce0d6ca4ec181e23cb2a
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:35 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:22 +0200

arm: Convert VFP hotplug notifiers to state machine

Straight forward conversion plus commentary why code which is executed
in hotplug callbacks needs to be invoked before installing them.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.713612993@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm/vfp/vfpmodule.c   | 28 +++++++++++++++++-----------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/arch/arm/vfp/vfpmodule.c b/arch/arm/vfp/vfpmodule.c
index 73085d3..da0b33d 100644
--- a/arch/arm/vfp/vfpmodule.c
+++ b/arch/arm/vfp/vfpmodule.c
@@ -643,19 +643,19 @@ int vfp_restore_user_hwstate(struct user_vfp __user *ufp,
  * hardware state at every thread switch.  We clear our held state when
  * a CPU has been killed, indicating that the VFP hardware doesn't contain
  * a threads VFP state.  When a CPU starts up, we re-enable access to the
- * VFP hardware.
- *
- * Both CPU_DYING and CPU_STARTING are called on the CPU which
+ * VFP hardware. The callbacks below are called on the CPU which
  * is being offlined/onlined.
  */
-static int vfp_hotplug(struct notifier_block *b, unsigned long action,
-	void *hcpu)
+static int vfp_dying_cpu(unsigned int cpu)
 {
-	if (action == CPU_DYING || action == CPU_DYING_FROZEN)
-		vfp_current_hw_state[(long)hcpu] = NULL;
-	else if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
-		vfp_enable(NULL);
-	return NOTIFY_OK;
+	vfp_force_reload(cpu, current_thread_info());
+	return 0;
+}
+
+static int vfp_starting_cpu(unsigned int unused)
+{
+	vfp_enable(NULL);
+	return 0;
 }
 
 void vfp_kmode_exception(void)
@@ -732,6 +732,10 @@ static int __init vfp_init(void)
 	unsigned int vfpsid;
 	unsigned int cpu_arch = cpu_architecture();
 
+	/*
+	 * Enable the access to the VFP on all online CPUs so the
+	 * following test on FPSID will succeed.
+	 */
 	if (cpu_arch >= CPU_ARCH_ARMv6)
 		on_each_cpu(vfp_enable, NULL, 1);
 
@@ -794,7 +798,9 @@ static int __init vfp_init(void)
 		VFP_arch = (vfpsid & FPSID_ARCH_MASK) >> FPSID_ARCH_BIT;
 	}
 
-	hotcpu_notifier(vfp_hotplug, 0);
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_VFP_STARTING,
+				  "AP_ARM_VFP_STARTING", vfp_starting_cpu,
+				  vfp_dying_cpu);
 
 	vfp_vector = vfp_support_entry;
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 3e719ba..4c63cb3 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -34,6 +34,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
+	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 35/67] arm/perf: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0034-arm-perf-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3771 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Straight forward conversion w/o bells and whistles.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/perf/arm_pmu.c       | 36 +++++++++++++++---------------------
 include/linux/cpuhotplug.h   |  1 +
 include/linux/perf/arm_pmu.h |  1 -
 3 files changed, 16 insertions(+), 22 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index 140436a..ae9fc6c 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -691,24 +691,15 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
  * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
  * junk values out of them.
  */
-static int cpu_pmu_notify(struct notifier_block *b, unsigned long action,
-			  void *hcpu)
+static int arm_perf_starting_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
-	struct arm_pmu *pmu = container_of(b, struct arm_pmu, hotplug_nb);
-
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
-
-	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-		return NOTIFY_DONE;
-
-	if (pmu->reset)
-		pmu->reset(pmu);
-	else
-		return NOTIFY_DONE;
-
-	return NOTIFY_OK;
+	if (!__oprofile_cpu_pmu)
+		return 0;
+	if (!cpumask_test_cpu(cpu, &__oprofile_cpu_pmu->supported_cpus))
+		return 0;
+	if (__oprofile_cpu_pmu->reset)
+		__oprofile_cpu_pmu->reset(__oprofile_cpu_pmu);
+	return 0;
 }
 
 #ifdef CONFIG_CPU_PM
@@ -819,8 +810,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	cpu_pmu->hotplug_nb.notifier_call = cpu_pmu_notify;
-	err = register_cpu_notifier(&cpu_pmu->hotplug_nb);
+	err = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					"AP_PERF_ARM_STARTING",
+					arm_perf_starting_cpu, NULL);
 	if (err)
 		goto out_hw_events;
 
@@ -858,7 +850,7 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
 out_hw_events:
 	free_percpu(cpu_hw_events);
 	return err;
@@ -867,7 +859,7 @@ out_hw_events:
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1027,6 +1019,8 @@ int arm_pmu_device_probe(struct platform_device *pdev,
 	if (ret)
 		goto out_destroy;
 
+	WARN(__oprofile_cpu_pmu, "%s(): missing PMU strucure for CPU-hotplug\n",
+	     __func__);
 	if (!__oprofile_cpu_pmu)
 		__oprofile_cpu_pmu = pmu;
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 2e227b6..365d615 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -36,6 +36,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
+	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index d28ac05..e6ed34e 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -109,7 +109,6 @@ struct arm_pmu {
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);
 	struct platform_device	*plat_device;
 	struct pmu_hw_events	__percpu *hw_events;
-	struct notifier_block	hotplug_nb;
 	struct notifier_block	cpu_pm_nb;
 };
 
-- 
2.8.1

Re: [patch V2 35/67] arm/perf: Convert to hotplug state machine

[Mark Rutland]

Hi,

On Wed, Jul 13, 2016 at 05:16:36PM +0000, Anna-Maria Gleixner wrote:
> From: Thomas Gleixner <tglx@linutronix.de>
> 
> Straight forward conversion w/o bells and whistles.
> 
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Will Deacon <will.deacon@arm.com>
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
> ---
>  drivers/perf/arm_pmu.c       | 36 +++++++++++++++---------------------
>  include/linux/cpuhotplug.h   |  1 +
>  include/linux/perf/arm_pmu.h |  1 -
>  3 files changed, 16 insertions(+), 22 deletions(-)
> 
> diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
> index 140436a..ae9fc6c 100644
> --- a/drivers/perf/arm_pmu.c
> +++ b/drivers/perf/arm_pmu.c
> @@ -691,24 +691,15 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
>   * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
>   * junk values out of them.
>   */
> -static int cpu_pmu_notify(struct notifier_block *b, unsigned long action,
> -			  void *hcpu)
> +static int arm_perf_starting_cpu(unsigned int cpu)
>  {
> -	int cpu = (unsigned long)hcpu;
> -	struct arm_pmu *pmu = container_of(b, struct arm_pmu, hotplug_nb);
> -
> -	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
> -		return NOTIFY_DONE;
> -
> -	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
> -		return NOTIFY_DONE;
> -
> -	if (pmu->reset)
> -		pmu->reset(pmu);
> -	else
> -		return NOTIFY_DONE;
> -
> -	return NOTIFY_OK;
> +	if (!__oprofile_cpu_pmu)
> +		return 0;
> +	if (!cpumask_test_cpu(cpu, &__oprofile_cpu_pmu->supported_cpus))
> +		return 0;
> +	if (__oprofile_cpu_pmu->reset)
> +		__oprofile_cpu_pmu->reset(__oprofile_cpu_pmu);
> +	return 0;
>  }

We may have multiple PMUs (e.g. two in big.LITTLE systems), and
__oprofile_cpu_pmu only contains one of these. So this conversion is not
correct.

We were relying on the notifier list implicitly containing a list of
those PMUs. It seems like we need an explicit list here.

We keep __oprofile_cpu_pmu around for legacy 32-bit users of OProfile
(on non-hetereogeneous systems), and that's all that the variable should
be used for.

Thanks,
Mark.

Re: [patch V2 35/67] arm/perf: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

* Mark Rutland | 2016-07-15 14:08:18 [+0100]:

>Hi,
Hi Mark,

>We may have multiple PMUs (e.g. two in big.LITTLE systems), and
>__oprofile_cpu_pmu only contains one of these. So this conversion is not
>correct.
>
>We were relying on the notifier list implicitly containing a list of
>those PMUs. It seems like we need an explicit list here.
>
>We keep __oprofile_cpu_pmu around for legacy 32-bit users of OProfile
>(on non-hetereogeneous systems), and that's all that the variable should
>be used for.

By the time I'e been looking there was only one node in the .dts. Now
you say it is not only possible but likely to have more than one node.

So what we need is a list which gets extended in cpu_pmu_init() instead
of using the global pointer we have now. Correct?

>Thanks,
>Mark.

Sebastian

Re: [patch V2 35/67] arm/perf: Convert to hotplug state machine

[Mark Rutland]

On Fri, Jul 15, 2016 at 05:26:30PM +0200, Sebastian Andrzej Siewior wrote:
> * Mark Rutland | 2016-07-15 14:08:18 [+0100]:
> 
> >Hi,
> Hi Mark,
> 
> >We may have multiple PMUs (e.g. two in big.LITTLE systems), and
> >__oprofile_cpu_pmu only contains one of these. So this conversion is not
> >correct.
> >
> >We were relying on the notifier list implicitly containing a list of
> >those PMUs. It seems like we need an explicit list here.
> >
> >We keep __oprofile_cpu_pmu around for legacy 32-bit users of OProfile
> >(on non-hetereogeneous systems), and that's all that the variable should
> >be used for.
> 
> By the time I'e been looking there was only one node in the .dts. Now
> you say it is not only possible but likely to have more than one node.

Yes. For example, arch/arm/boot/dts/vexpress-v2p-ca15_a7.dts is a
big.LITTLE system with a Cortex-A7 PMU and a Cortex-A15 PMU.

> So what we need is a list which gets extended in cpu_pmu_init() instead
> of using the global pointer we have now. Correct?

Yes, that sounds right to me, with matching removal in cpu_pmu_destroy.

Thanks,
Mark.

[tip:smp/hotplug] arm/perf: Convert to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  7d88eb695a1f5f67820e02999b949d5cfa080442
Gitweb:     http://git.kernel.org/tip/7d88eb695a1f5f67820e02999b949d5cfa080442
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:36 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:23 +0200

arm/perf: Convert to hotplug state machine

Straight forward conversion w/o bells and whistles.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.794097159@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/perf/arm_pmu.c       | 36 +++++++++++++++---------------------
 include/linux/cpuhotplug.h   |  1 +
 include/linux/perf/arm_pmu.h |  1 -
 3 files changed, 16 insertions(+), 22 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index 140436a..ae9fc6c 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -691,24 +691,15 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
  * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
  * junk values out of them.
  */
-static int cpu_pmu_notify(struct notifier_block *b, unsigned long action,
-			  void *hcpu)
+static int arm_perf_starting_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
-	struct arm_pmu *pmu = container_of(b, struct arm_pmu, hotplug_nb);
-
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
-
-	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-		return NOTIFY_DONE;
-
-	if (pmu->reset)
-		pmu->reset(pmu);
-	else
-		return NOTIFY_DONE;
-
-	return NOTIFY_OK;
+	if (!__oprofile_cpu_pmu)
+		return 0;
+	if (!cpumask_test_cpu(cpu, &__oprofile_cpu_pmu->supported_cpus))
+		return 0;
+	if (__oprofile_cpu_pmu->reset)
+		__oprofile_cpu_pmu->reset(__oprofile_cpu_pmu);
+	return 0;
 }
 
 #ifdef CONFIG_CPU_PM
@@ -819,8 +810,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	cpu_pmu->hotplug_nb.notifier_call = cpu_pmu_notify;
-	err = register_cpu_notifier(&cpu_pmu->hotplug_nb);
+	err = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					"AP_PERF_ARM_STARTING",
+					arm_perf_starting_cpu, NULL);
 	if (err)
 		goto out_hw_events;
 
@@ -858,7 +850,7 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
 out_hw_events:
 	free_percpu(cpu_hw_events);
 	return err;
@@ -867,7 +859,7 @@ out_hw_events:
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1027,6 +1019,8 @@ int arm_pmu_device_probe(struct platform_device *pdev,
 	if (ret)
 		goto out_destroy;
 
+	WARN(__oprofile_cpu_pmu, "%s(): missing PMU strucure for CPU-hotplug\n",
+	     __func__);
 	if (!__oprofile_cpu_pmu)
 		__oprofile_cpu_pmu = pmu;
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4c63cb3..a5b6a65 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
+	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index d28ac05..e6ed34e 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -109,7 +109,6 @@ struct arm_pmu {
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);
 	struct platform_device	*plat_device;
 	struct pmu_hw_events	__percpu *hw_events;
-	struct notifier_block	hotplug_nb;
 	struct notifier_block	cpu_pm_nb;
 };
 

Re: [tip:smp/hotplug] arm/perf: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

On 2016-07-19 00:16:15 [-0700], tip-bot for Thomas Gleixner wrote:
> Commit-ID:  7d88eb695a1f5f67820e02999b949d5cfa080442
> Gitweb:     http://git.kernel.org/tip/7d88eb695a1f5f67820e02999b949d5cfa080442
> Author:     Thomas Gleixner <tglx@linutronix.de>
> AuthorDate: Wed, 13 Jul 2016 17:16:36 +0000
> Committer:  Ingo Molnar <mingo@kernel.org>
> CommitDate: Fri, 15 Jul 2016 10:40:23 +0200
> 

Ingo please drop this. I have a new version almost done (Mark asked for
it).

Sebastian

Re: [tip:smp/hotplug] arm/perf: Convert to hotplug state machine

[Ingo Molnar]

* Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote:

> On 2016-07-19 00:16:15 [-0700], tip-bot for Thomas Gleixner wrote:
> > Commit-ID:  7d88eb695a1f5f67820e02999b949d5cfa080442
> > Gitweb:     http://git.kernel.org/tip/7d88eb695a1f5f67820e02999b949d5cfa080442
> > Author:     Thomas Gleixner <tglx@linutronix.de>
> > AuthorDate: Wed, 13 Jul 2016 17:16:36 +0000
> > Committer:  Ingo Molnar <mingo@kernel.org>
> > CommitDate: Fri, 15 Jul 2016 10:40:23 +0200
> > 
> 
> Ingo please drop this. I have a new version almost done (Mark asked for
> it).

Please send me an updated patch ASAP.

Thanks,

	Ingo

[PATCH v2] arm/perf: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

From: Thomas Gleixner <tglx@linutronix.de>

Straight forward conversion w/o bells and whistles.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
[bigeasy: use arm_pmu_list to handle multiple  PMUs in the system]
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/perf/arm_pmu.c       | 59 +++++++++++++++++++++++++++-----------------
 include/linux/cpuhotplug.h   |  1 +
 include/linux/perf/arm_pmu.h |  2 +-
 3 files changed, 39 insertions(+), 23 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index 140436a046c0..b7a64c0ec673 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -685,30 +685,29 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
+static DEFINE_MUTEX(arm_pmu_mutex);
+static LIST_HEAD(arm_pmu_list);
+
 /*
  * PMU hardware loses all context when a CPU goes offline.
  * When a CPU is hotplugged back in, since some hardware registers are
  * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
  * junk values out of them.
  */
-static int cpu_pmu_notify(struct notifier_block *b, unsigned long action,
-			  void *hcpu)
+static int arm_perf_starting_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
-	struct arm_pmu *pmu = container_of(b, struct arm_pmu, hotplug_nb);
+	struct arm_pmu *pmu;
 
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
+	mutex_lock(&arm_pmu_mutex);
+	list_for_each_entry(pmu, &arm_pmu_list, entry) {
 
-	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-		return NOTIFY_DONE;
-
-	if (pmu->reset)
-		pmu->reset(pmu);
-	else
-		return NOTIFY_DONE;
-
-	return NOTIFY_OK;
+		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+			continue;
+		if (pmu->reset)
+			pmu->reset(pmu);
+	}
+	mutex_unlock(&arm_pmu_mutex);
+	return 0;
 }
 
 #ifdef CONFIG_CPU_PM
@@ -819,10 +818,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	cpu_pmu->hotplug_nb.notifier_call = cpu_pmu_notify;
-	err = register_cpu_notifier(&cpu_pmu->hotplug_nb);
-	if (err)
-		goto out_hw_events;
+	mutex_lock(&arm_pmu_mutex);
+	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
+	mutex_unlock(&arm_pmu_mutex);
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -858,8 +856,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
-out_hw_events:
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -867,7 +866,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1044,3 +1045,17 @@ int arm_pmu_device_probe(struct platform_device *pdev,
 	kfree(pmu);
 	return ret;
 }
+
+static int arm_pmu_hp_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					"AP_PERF_ARM_STARTING",
+					arm_perf_starting_cpu, NULL);
+	if (ret)
+		pr_err("CPU hotplug notifier for ARM PMU could not be "
+		       "registered: %d\n", ret);
+	return ret;
+}
+subsys_initcall(arm_pmu_hp_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4c63cb30aee6..a5b6a6526af8 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
+	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index d28ac05c7f92..e18843809eec 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -109,7 +109,7 @@ struct arm_pmu {
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);
 	struct platform_device	*plat_device;
 	struct pmu_hw_events	__percpu *hw_events;
-	struct notifier_block	hotplug_nb;
+	struct list_head	entry;
 	struct notifier_block	cpu_pm_nb;
 };
 
-- 
2.8.1

Re: [PATCH v2] arm/perf: Convert to hotplug state machine

[Mark Rutland]

Hi,

On Tue, Jul 19, 2016 at 11:02:29AM +0200, Sebastian Andrzej Siewior wrote:
> +static int arm_pmu_hp_init(void)
> +{
> +	int ret;
> +
> +	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
> +					"AP_PERF_ARM_STARTING",
> +					arm_perf_starting_cpu, NULL);
> +	if (ret)
> +		pr_err("CPU hotplug notifier for ARM PMU could not be "
> +		       "registered: %d\n", ret);

Nit: Per Documentation/CodingStyle strings should always be on one line,
even if greater than 80 chars, so they're greppable. I assume that can
be fixed up when this is picked.

Otherwise, v2 looks correct to me. FWIW:

Acked-by: Mark Rutland <mark.rutland@arm.com>

Thanks,
Mark.

[PATCH v3] arm/perf: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

From: Thomas Gleixner <tglx@linutronix.de>

Straight forward conversion w/o bells and whistles.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
[bigeasy@l.de: use arm_pmu_list to handle multiple PMUs in the system]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
v2…v3: >80 chars line for per_err, added Mark's Acked-by
v1…v2: use arm_pmu_list

 drivers/perf/arm_pmu.c       | 59 +++++++++++++++++++++++++++-----------------
 include/linux/cpuhotplug.h   |  1 +
 include/linux/perf/arm_pmu.h |  2 +-
 3 files changed, 39 insertions(+), 23 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index 140436a046c0..f6ab4f7f75bf 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -685,30 +685,29 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
+static DEFINE_MUTEX(arm_pmu_mutex);
+static LIST_HEAD(arm_pmu_list);
+
 /*
  * PMU hardware loses all context when a CPU goes offline.
  * When a CPU is hotplugged back in, since some hardware registers are
  * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
  * junk values out of them.
  */
-static int cpu_pmu_notify(struct notifier_block *b, unsigned long action,
-			  void *hcpu)
+static int arm_perf_starting_cpu(unsigned int cpu)
 {
-	int cpu = (unsigned long)hcpu;
-	struct arm_pmu *pmu = container_of(b, struct arm_pmu, hotplug_nb);
+	struct arm_pmu *pmu;
 
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
+	mutex_lock(&arm_pmu_mutex);
+	list_for_each_entry(pmu, &arm_pmu_list, entry) {
 
-	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-		return NOTIFY_DONE;
-
-	if (pmu->reset)
-		pmu->reset(pmu);
-	else
-		return NOTIFY_DONE;
-
-	return NOTIFY_OK;
+		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+			continue;
+		if (pmu->reset)
+			pmu->reset(pmu);
+	}
+	mutex_unlock(&arm_pmu_mutex);
+	return 0;
 }
 
 #ifdef CONFIG_CPU_PM
@@ -819,10 +818,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	cpu_pmu->hotplug_nb.notifier_call = cpu_pmu_notify;
-	err = register_cpu_notifier(&cpu_pmu->hotplug_nb);
-	if (err)
-		goto out_hw_events;
+	mutex_lock(&arm_pmu_mutex);
+	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
+	mutex_unlock(&arm_pmu_mutex);
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -858,8 +856,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
-out_hw_events:
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -867,7 +866,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1044,3 +1045,17 @@ int arm_pmu_device_probe(struct platform_device *pdev,
 	kfree(pmu);
 	return ret;
 }
+
+static int arm_pmu_hp_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					"AP_PERF_ARM_STARTING",
+					arm_perf_starting_cpu, NULL);
+	if (ret)
+		pr_err("CPU hotplug notifier for ARM PMU could not be registered: %d\n",
+		       ret);
+	return ret;
+}
+subsys_initcall(arm_pmu_hp_init);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4c63cb30aee6..a5b6a6526af8 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
+	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index d28ac05c7f92..e18843809eec 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -109,7 +109,7 @@ struct arm_pmu {
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);
 	struct platform_device	*plat_device;
 	struct pmu_hw_events	__percpu *hw_events;
-	struct notifier_block	hotplug_nb;
+	struct list_head	entry;
 	struct notifier_block	cpu_pm_nb;
 };
 
-- 
2.8.1

Re: [PATCH v3] arm/perf: Convert to hotplug state machine

[Ingo Molnar]

* Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote:

> From: Thomas Gleixner <tglx@linutronix.de>
> 
> Straight forward conversion w/o bells and whistles.
> 
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Acked-by: Mark Rutland <mark.rutland@arm.com>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Will Deacon <will.deacon@arm.com>
> Cc: rt@linutronix.de
> [bigeasy@l.de: use arm_pmu_list to handle multiple PMUs in the system]
> Signed-off-by: Ingo Molnar <mingo@kernel.org>
> ---
> v2…v3: >80 chars line for per_err, added Mark's Acked-by
> v1…v2: use arm_pmu_list

Thanks guys. I converted this commit into a fix patch with a proper changelog and 
added Sebastian's SOB and Mark's Acked-by (there's no changes to the resulting 
code) - this will better document the reasoning behind the introduction of 
'arm_pmu_list' and also avoids having to rebase 30+ commits.

The tip-bot will post the patch to this thread soon, please double check the 
result and let me know if anything is amiss!

Thanks,

	Ingo

[tip:smp/hotplug] arm/perf: Fix hotplug state machine conversion

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  37b502f121adab26ccc2769c3063f0e1272be7de
Gitweb:     http://git.kernel.org/tip/37b502f121adab26ccc2769c3063f0e1272be7de
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 20 Jul 2016 09:51:11 +0200
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 20 Jul 2016 09:57:34 +0200

arm/perf: Fix hotplug state machine conversion

Mark Rutland pointed out that this commit is incomplete:

  7d88eb695a1f ("arm/perf: Convert to hotplug state machine")

The problem is that:

 > We may have multiple PMUs (e.g. two in big.LITTLE systems), and
 > __oprofile_cpu_pmu only contains one of these. So this conversion is not
 > correct.
 >
 > We were relying on the notifier list implicitly containing a list of
 > those PMUs. It seems like we need an explicit list here.
 >
 > We keep __oprofile_cpu_pmu around for legacy 32-bit users of OProfile
 > (on non-hetereogeneous systems), and that's all that the variable should
 > be used for.

Introduce arm_pmu_list to correctly handle multiple PMUs in the system.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Cc: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-tip-commits@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160719111733.GA22911@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/perf/arm_pmu.c       | 53 +++++++++++++++++++++++++++++++-------------
 include/linux/perf/arm_pmu.h |  1 +
 2 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index ae9fc6c..f6ab4f7 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -685,6 +685,9 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
+static DEFINE_MUTEX(arm_pmu_mutex);
+static LIST_HEAD(arm_pmu_list);
+
 /*
  * PMU hardware loses all context when a CPU goes offline.
  * When a CPU is hotplugged back in, since some hardware registers are
@@ -693,12 +696,17 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
  */
 static int arm_perf_starting_cpu(unsigned int cpu)
 {
-	if (!__oprofile_cpu_pmu)
-		return 0;
-	if (!cpumask_test_cpu(cpu, &__oprofile_cpu_pmu->supported_cpus))
-		return 0;
-	if (__oprofile_cpu_pmu->reset)
-		__oprofile_cpu_pmu->reset(__oprofile_cpu_pmu);
+	struct arm_pmu *pmu;
+
+	mutex_lock(&arm_pmu_mutex);
+	list_for_each_entry(pmu, &arm_pmu_list, entry) {
+
+		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+			continue;
+		if (pmu->reset)
+			pmu->reset(pmu);
+	}
+	mutex_unlock(&arm_pmu_mutex);
 	return 0;
 }
 
@@ -810,11 +818,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	err = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
-					"AP_PERF_ARM_STARTING",
-					arm_perf_starting_cpu, NULL);
-	if (err)
-		goto out_hw_events;
+	mutex_lock(&arm_pmu_mutex);
+	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
+	mutex_unlock(&arm_pmu_mutex);
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -850,8 +856,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
-out_hw_events:
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -859,7 +866,9 @@ out_hw_events:
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_STARTING);
+	mutex_lock(&arm_pmu_mutex);
+	list_del(&cpu_pmu->entry);
+	mutex_unlock(&arm_pmu_mutex);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1019,8 +1028,6 @@ int arm_pmu_device_probe(struct platform_device *pdev,
 	if (ret)
 		goto out_destroy;
 
-	WARN(__oprofile_cpu_pmu, "%s(): missing PMU strucure for CPU-hotplug\n",
-	     __func__);
 	if (!__oprofile_cpu_pmu)
 		__oprofile_cpu_pmu = pmu;
 
@@ -1038,3 +1045,17 @@ out_free:
 	kfree(pmu);
 	return ret;
 }
+
+static int arm_pmu_hp_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					"AP_PERF_ARM_STARTING",
+					arm_perf_starting_cpu, NULL);
+	if (ret)
+		pr_err("CPU hotplug notifier for ARM PMU could not be registered: %d\n",
+		       ret);
+	return ret;
+}
+subsys_initcall(arm_pmu_hp_init);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index e6ed34e..e188438 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -109,6 +109,7 @@ struct arm_pmu {
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);
 	struct platform_device	*plat_device;
 	struct pmu_hw_events	__percpu *hw_events;
+	struct list_head	entry;
 	struct notifier_block	cpu_pm_nb;
 };
 

[patch V2 36/67] virt: Convert kvm hotplug to state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0035-virt-Convert-kvm-hotplug-to-state-machine.patch --]
[-- Type: text/plain, Size: 3813 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Install the callbacks via the state machine. The core won't invoke the
callbacks on already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: kvm@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/kvm_main.c        | 32 ++++++++------------------------
 2 files changed, 9 insertions(+), 24 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 365d615..1d71f38 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -37,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 48bd520..c1d6cf5 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -3144,12 +3144,13 @@ static void hardware_enable_nolock(void *junk)
 	}
 }
 
-static void hardware_enable(void)
+static int kvm_starting_cpu(unsigned int cpu)
 {
 	raw_spin_lock(&kvm_count_lock);
 	if (kvm_usage_count)
 		hardware_enable_nolock(NULL);
 	raw_spin_unlock(&kvm_count_lock);
+	return 0;
 }
 
 static void hardware_disable_nolock(void *junk)
@@ -3162,12 +3163,13 @@ static void hardware_disable_nolock(void *junk)
 	kvm_arch_hardware_disable();
 }
 
-static void hardware_disable(void)
+static int kvm_dying_cpu(unsigned int cpu)
 {
 	raw_spin_lock(&kvm_count_lock);
 	if (kvm_usage_count)
 		hardware_disable_nolock(NULL);
 	raw_spin_unlock(&kvm_count_lock);
+	return 0;
 }
 
 static void hardware_disable_all_nolock(void)
@@ -3208,21 +3210,6 @@ static int hardware_enable_all(void)
 	return r;
 }
 
-static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
-			   void *v)
-{
-	val &= ~CPU_TASKS_FROZEN;
-	switch (val) {
-	case CPU_DYING:
-		hardware_disable();
-		break;
-	case CPU_STARTING:
-		hardware_enable();
-		break;
-	}
-	return NOTIFY_OK;
-}
-
 static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
 		      void *v)
 {
@@ -3489,10 +3476,6 @@ int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
 	return r;
 }
 
-static struct notifier_block kvm_cpu_notifier = {
-	.notifier_call = kvm_cpu_hotplug,
-};
-
 static int kvm_debugfs_open(struct inode *inode, struct file *file,
 			   int (*get)(void *, u64 *), int (*set)(void *, u64),
 			   const char *fmt)
@@ -3743,7 +3726,8 @@ int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
 			goto out_free_1;
 	}
 
-	r = register_cpu_notifier(&kvm_cpu_notifier);
+	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "AP_KVM_STARTING",
+				      kvm_starting_cpu, kvm_dying_cpu);
 	if (r)
 		goto out_free_2;
 	register_reboot_notifier(&kvm_reboot_notifier);
@@ -3797,7 +3781,7 @@ out_free:
 	kmem_cache_destroy(kvm_vcpu_cache);
 out_free_3:
 	unregister_reboot_notifier(&kvm_reboot_notifier);
-	unregister_cpu_notifier(&kvm_cpu_notifier);
+	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
 out_free_2:
 out_free_1:
 	kvm_arch_hardware_unsetup();
@@ -3820,7 +3804,7 @@ void kvm_exit(void)
 	kvm_async_pf_deinit();
 	unregister_syscore_ops(&kvm_syscore_ops);
 	unregister_reboot_notifier(&kvm_reboot_notifier);
-	unregister_cpu_notifier(&kvm_cpu_notifier);
+	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
 	on_each_cpu(hardware_disable_nolock, NULL, 1);
 	kvm_arch_hardware_unsetup();
 	kvm_arch_exit();
-- 
2.8.1

[tip:smp/hotplug] virt: Convert kvm hotplug to state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  8c18b2d2d0881b116ea52498c6d624c0a71e1fdc
Gitweb:     http://git.kernel.org/tip/8c18b2d2d0881b116ea52498c6d624c0a71e1fdc
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:37 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:23 +0200

virt: Convert kvm hotplug to state machine

Install the callbacks via the state machine. The core won't invoke the
callbacks on already online CPUs.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: kvm@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.886159080@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/kvm_main.c        | 32 ++++++++------------------------
 2 files changed, 9 insertions(+), 24 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index a5b6a65..5423997 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -36,6 +36,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 48bd520..c1d6cf5 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -3144,12 +3144,13 @@ static void hardware_enable_nolock(void *junk)
 	}
 }
 
-static void hardware_enable(void)
+static int kvm_starting_cpu(unsigned int cpu)
 {
 	raw_spin_lock(&kvm_count_lock);
 	if (kvm_usage_count)
 		hardware_enable_nolock(NULL);
 	raw_spin_unlock(&kvm_count_lock);
+	return 0;
 }
 
 static void hardware_disable_nolock(void *junk)
@@ -3162,12 +3163,13 @@ static void hardware_disable_nolock(void *junk)
 	kvm_arch_hardware_disable();
 }
 
-static void hardware_disable(void)
+static int kvm_dying_cpu(unsigned int cpu)
 {
 	raw_spin_lock(&kvm_count_lock);
 	if (kvm_usage_count)
 		hardware_disable_nolock(NULL);
 	raw_spin_unlock(&kvm_count_lock);
+	return 0;
 }
 
 static void hardware_disable_all_nolock(void)
@@ -3208,21 +3210,6 @@ static int hardware_enable_all(void)
 	return r;
 }
 
-static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
-			   void *v)
-{
-	val &= ~CPU_TASKS_FROZEN;
-	switch (val) {
-	case CPU_DYING:
-		hardware_disable();
-		break;
-	case CPU_STARTING:
-		hardware_enable();
-		break;
-	}
-	return NOTIFY_OK;
-}
-
 static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
 		      void *v)
 {
@@ -3489,10 +3476,6 @@ int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
 	return r;
 }
 
-static struct notifier_block kvm_cpu_notifier = {
-	.notifier_call = kvm_cpu_hotplug,
-};
-
 static int kvm_debugfs_open(struct inode *inode, struct file *file,
 			   int (*get)(void *, u64 *), int (*set)(void *, u64),
 			   const char *fmt)
@@ -3743,7 +3726,8 @@ int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
 			goto out_free_1;
 	}
 
-	r = register_cpu_notifier(&kvm_cpu_notifier);
+	r = cpuhp_setup_state_nocalls(CPUHP_AP_KVM_STARTING, "AP_KVM_STARTING",
+				      kvm_starting_cpu, kvm_dying_cpu);
 	if (r)
 		goto out_free_2;
 	register_reboot_notifier(&kvm_reboot_notifier);
@@ -3797,7 +3781,7 @@ out_free:
 	kmem_cache_destroy(kvm_vcpu_cache);
 out_free_3:
 	unregister_reboot_notifier(&kvm_reboot_notifier);
-	unregister_cpu_notifier(&kvm_cpu_notifier);
+	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
 out_free_2:
 out_free_1:
 	kvm_arch_hardware_unsetup();
@@ -3820,7 +3804,7 @@ void kvm_exit(void)
 	kvm_async_pf_deinit();
 	unregister_syscore_ops(&kvm_syscore_ops);
 	unregister_reboot_notifier(&kvm_reboot_notifier);
-	unregister_cpu_notifier(&kvm_cpu_notifier);
+	cpuhp_remove_state_nocalls(CPUHP_AP_KVM_STARTING);
 	on_each_cpu(hardware_disable_nolock, NULL, 1);
 	kvm_arch_hardware_unsetup();
 	kvm_arch_exit();

[patch V2 37/67] ACPI/processor: Avoid STARTING/DYING actions in a more logical way

[Anna-Maria Gleixner]

[-- Attachment #1: 0036-ACPI-processor-Avoid-STARTING-DYING-actions-in-a-mor.patch --]
[-- Type: text/plain, Size: 1816 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

As part of the hotplug cleanup, the CPU_STARTING/DYING actions are going
away soon.  This driver needlessly uses those two macro, and so this patch
replaces that code with something more sensible.

Commit:

  8da8373447d6a57a5a9f55233d35beb15d92d0d2 ("ACPI / processor: Fix STARTING/DYING action in acpi_cpu_soft_notify()")

added checks for those two actions, because the notification callback can
sleep, causing a hung CPU. This patch instead checks for the ONLINE/DEAD
actions, which are the ones that are handled by the driver in the first
place.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Len Brown <lenb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-acpi@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/acpi/processor_driver.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/processor_driver.c b/drivers/acpi/processor_driver.c
index d2fa8cb..eecdb19 100644
--- a/drivers/acpi/processor_driver.c
+++ b/drivers/acpi/processor_driver.c
@@ -118,12 +118,13 @@ static int acpi_cpu_soft_notify(struct notifier_block *nfb,
 	struct acpi_device *device;
 	action &= ~CPU_TASKS_FROZEN;
 
-	/*
-	 * CPU_STARTING and CPU_DYING must not sleep. Return here since
-	 * acpi_bus_get_device() may sleep.
-	 */
-	if (action == CPU_STARTING || action == CPU_DYING)
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_DEAD:
+		break;
+	default:
 		return NOTIFY_DONE;
+	}
 
 	if (!pr || acpi_bus_get_device(pr->handle, &device))
 		return NOTIFY_DONE;
-- 
2.8.1

[tip:smp/hotplug] ACPI/processor: Avoid STARTING/DYING actions in a more logical way

[tip-bot for Richard Cochran]

Commit-ID:  e1272f541c6ad476426ef24b56b4e1d3403db513
Gitweb:     http://git.kernel.org/tip/e1272f541c6ad476426ef24b56b4e1d3403db513
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:38 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:24 +0200

ACPI/processor: Avoid STARTING/DYING actions in a more logical way

As part of the hotplug cleanup, the CPU_STARTING/DYING actions are going
away soon.  This driver needlessly uses those two macro, and so this patch
replaces that code with something more sensible.

Commit:

  8da8373447d6a57a5a9f55233d35beb15d92d0d2 ("ACPI / processor: Fix STARTING/DYING action in acpi_cpu_soft_notify()")

added checks for those two actions, because the notification callback can
sleep, causing a hung CPU. This patch instead checks for the ONLINE/DEAD
actions, which are the ones that are handled by the driver in the first
place.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Len Brown <lenb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-acpi@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153335.964962885@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/acpi/processor_driver.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/processor_driver.c b/drivers/acpi/processor_driver.c
index d2fa8cb..eecdb19 100644
--- a/drivers/acpi/processor_driver.c
+++ b/drivers/acpi/processor_driver.c
@@ -118,12 +118,13 @@ static int acpi_cpu_soft_notify(struct notifier_block *nfb,
 	struct acpi_device *device;
 	action &= ~CPU_TASKS_FROZEN;
 
-	/*
-	 * CPU_STARTING and CPU_DYING must not sleep. Return here since
-	 * acpi_bus_get_device() may sleep.
-	 */
-	if (action == CPU_STARTING || action == CPU_DYING)
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_DEAD:
+		break;
+	default:
 		return NOTIFY_DONE;
+	}
 
 	if (!pr || acpi_bus_get_device(pr->handle, &device))
 		return NOTIFY_DONE;

[patch V2 38/67] clocksource/arm_arch_timer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0037-clocksource-arm_arch_timer-Convert-to-hotplug-state-.patch --]
[-- Type: text/plain, Size: 4025 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/arm_arch_timer.c | 54 +++++++++++++++++-------------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 27 insertions(+), 28 deletions(-)

diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
index 5effd30..28bce3f 100644
--- a/drivers/clocksource/arm_arch_timer.c
+++ b/drivers/clocksource/arm_arch_timer.c
@@ -370,8 +370,10 @@ static bool arch_timer_has_nonsecure_ppi(void)
 		arch_timer_ppi[PHYS_NONSECURE_PPI]);
 }
 
-static int arch_timer_setup(struct clock_event_device *clk)
+static int arch_timer_starting_cpu(unsigned int cpu)
 {
+	struct clock_event_device *clk = this_cpu_ptr(arch_timer_evt);
+
 	__arch_timer_setup(ARCH_CP15_TIMER, clk);
 
 	enable_percpu_irq(arch_timer_ppi[arch_timer_uses_ppi], 0);
@@ -527,29 +529,14 @@ static void arch_timer_stop(struct clock_event_device *clk)
 	clk->set_state_shutdown(clk);
 }
 
-static int arch_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
+static int arch_timer_dying_cpu(unsigned int cpu)
 {
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		arch_timer_setup(this_cpu_ptr(arch_timer_evt));
-		break;
-	case CPU_DYING:
-		arch_timer_stop(this_cpu_ptr(arch_timer_evt));
-		break;
-	}
+	struct clock_event_device *clk = this_cpu_ptr(arch_timer_evt);
 
-	return NOTIFY_OK;
+	arch_timer_stop(clk);
+	return 0;
 }
 
-static struct notifier_block arch_timer_cpu_nb = {
-	.notifier_call = arch_timer_cpu_notify,
-};
-
 #ifdef CONFIG_CPU_PM
 static unsigned int saved_cntkctl;
 static int arch_timer_cpu_pm_notify(struct notifier_block *self,
@@ -570,11 +557,21 @@ static int __init arch_timer_cpu_pm_init(void)
 {
 	return cpu_pm_register_notifier(&arch_timer_cpu_pm_notifier);
 }
+
+static void __init arch_timer_cpu_pm_deinit(void)
+{
+	WARN_ON(cpu_pm_unregister_notifier(&arch_timer_cpu_pm_notifier));
+}
+
 #else
 static int __init arch_timer_cpu_pm_init(void)
 {
 	return 0;
 }
+
+static void __init arch_timer_cpu_pm_deinit(void)
+{
+}
 #endif
 
 static int __init arch_timer_register(void)
@@ -621,22 +618,23 @@ static int __init arch_timer_register(void)
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&arch_timer_cpu_nb);
-	if (err)
-		goto out_free_irq;
-
 	err = arch_timer_cpu_pm_init();
 	if (err)
 		goto out_unreg_notify;
 
-	/* Immediately configure the timer on the boot CPU */
-	arch_timer_setup(this_cpu_ptr(arch_timer_evt));
 
+	/* Register and immediately configure the timer on the boot CPU */
+	err = cpuhp_setup_state(CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+				"AP_ARM_ARCH_TIMER_STARTING",
+				arch_timer_starting_cpu, arch_timer_dying_cpu);
+	if (err)
+		goto out_unreg_cpupm;
 	return 0;
 
+out_unreg_cpupm:
+	arch_timer_cpu_pm_deinit();
+
 out_unreg_notify:
-	unregister_cpu_notifier(&arch_timer_cpu_nb);
-out_free_irq:
 	free_percpu_irq(arch_timer_ppi[arch_timer_uses_ppi], arch_timer_evt);
 	if (arch_timer_has_nonsecure_ppi())
 		free_percpu_irq(arch_timer_ppi[PHYS_NONSECURE_PPI],
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 1d71f38..672f690 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -37,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] clocksource/arm_arch_timer: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  7e86e8bd8dd67649d176e08d8dfb90039f0a1e98
Gitweb:     http://git.kernel.org/tip/7e86e8bd8dd67649d176e08d8dfb90039f0a1e98
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:39 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:24 +0200

clocksource/arm_arch_timer: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.048259040@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/arm_arch_timer.c | 54 +++++++++++++++++-------------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 27 insertions(+), 28 deletions(-)

diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
index 5effd30..28bce3f 100644
--- a/drivers/clocksource/arm_arch_timer.c
+++ b/drivers/clocksource/arm_arch_timer.c
@@ -370,8 +370,10 @@ static bool arch_timer_has_nonsecure_ppi(void)
 		arch_timer_ppi[PHYS_NONSECURE_PPI]);
 }
 
-static int arch_timer_setup(struct clock_event_device *clk)
+static int arch_timer_starting_cpu(unsigned int cpu)
 {
+	struct clock_event_device *clk = this_cpu_ptr(arch_timer_evt);
+
 	__arch_timer_setup(ARCH_CP15_TIMER, clk);
 
 	enable_percpu_irq(arch_timer_ppi[arch_timer_uses_ppi], 0);
@@ -527,29 +529,14 @@ static void arch_timer_stop(struct clock_event_device *clk)
 	clk->set_state_shutdown(clk);
 }
 
-static int arch_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
+static int arch_timer_dying_cpu(unsigned int cpu)
 {
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		arch_timer_setup(this_cpu_ptr(arch_timer_evt));
-		break;
-	case CPU_DYING:
-		arch_timer_stop(this_cpu_ptr(arch_timer_evt));
-		break;
-	}
+	struct clock_event_device *clk = this_cpu_ptr(arch_timer_evt);
 
-	return NOTIFY_OK;
+	arch_timer_stop(clk);
+	return 0;
 }
 
-static struct notifier_block arch_timer_cpu_nb = {
-	.notifier_call = arch_timer_cpu_notify,
-};
-
 #ifdef CONFIG_CPU_PM
 static unsigned int saved_cntkctl;
 static int arch_timer_cpu_pm_notify(struct notifier_block *self,
@@ -570,11 +557,21 @@ static int __init arch_timer_cpu_pm_init(void)
 {
 	return cpu_pm_register_notifier(&arch_timer_cpu_pm_notifier);
 }
+
+static void __init arch_timer_cpu_pm_deinit(void)
+{
+	WARN_ON(cpu_pm_unregister_notifier(&arch_timer_cpu_pm_notifier));
+}
+
 #else
 static int __init arch_timer_cpu_pm_init(void)
 {
 	return 0;
 }
+
+static void __init arch_timer_cpu_pm_deinit(void)
+{
+}
 #endif
 
 static int __init arch_timer_register(void)
@@ -621,22 +618,23 @@ static int __init arch_timer_register(void)
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&arch_timer_cpu_nb);
-	if (err)
-		goto out_free_irq;
-
 	err = arch_timer_cpu_pm_init();
 	if (err)
 		goto out_unreg_notify;
 
-	/* Immediately configure the timer on the boot CPU */
-	arch_timer_setup(this_cpu_ptr(arch_timer_evt));
 
+	/* Register and immediately configure the timer on the boot CPU */
+	err = cpuhp_setup_state(CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+				"AP_ARM_ARCH_TIMER_STARTING",
+				arch_timer_starting_cpu, arch_timer_dying_cpu);
+	if (err)
+		goto out_unreg_cpupm;
 	return 0;
 
+out_unreg_cpupm:
+	arch_timer_cpu_pm_deinit();
+
 out_unreg_notify:
-	unregister_cpu_notifier(&arch_timer_cpu_nb);
-out_free_irq:
 	free_percpu_irq(arch_timer_ppi[arch_timer_uses_ppi], arch_timer_evt);
 	if (arch_timer_has_nonsecure_ppi())
 		free_percpu_irq(arch_timer_ppi[PHYS_NONSECURE_PPI],
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 5423997..f0524cd 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -36,6 +36,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 39/67] clocksource/dummy_timer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0038-clocksource-dummy_timer-Convert-to-hotplug-state-mac.patch --]
[-- Type: text/plain, Size: 2722 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/dummy_timer.c | 36 ++++++------------------------------
 include/linux/cpuhotplug.h        |  1 +
 2 files changed, 7 insertions(+), 30 deletions(-)

diff --git a/drivers/clocksource/dummy_timer.c b/drivers/clocksource/dummy_timer.c
index 776b6c8..89f1c2e 100644
--- a/drivers/clocksource/dummy_timer.c
+++ b/drivers/clocksource/dummy_timer.c
@@ -16,10 +16,9 @@
 
 static DEFINE_PER_CPU(struct clock_event_device, dummy_timer_evt);
 
-static void dummy_timer_setup(void)
+static int dummy_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
-	struct clock_event_device *evt = raw_cpu_ptr(&dummy_timer_evt);
+	struct clock_event_device *evt = per_cpu_ptr(&dummy_timer_evt, cpu);
 
 	evt->name	= "dummy_timer";
 	evt->features	= CLOCK_EVT_FEAT_PERIODIC |
@@ -29,36 +28,13 @@ static void dummy_timer_setup(void)
 	evt->cpumask	= cpumask_of(cpu);
 
 	clockevents_register_device(evt);
+	return 0;
 }
 
-static int dummy_timer_cpu_notify(struct notifier_block *self,
-				      unsigned long action, void *hcpu)
-{
-	if ((action & ~CPU_TASKS_FROZEN) == CPU_STARTING)
-		dummy_timer_setup();
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block dummy_timer_cpu_nb = {
-	.notifier_call = dummy_timer_cpu_notify,
-};
-
 static int __init dummy_timer_register(void)
 {
-	int err = 0;
-
-	cpu_notifier_register_begin();
-	err = __register_cpu_notifier(&dummy_timer_cpu_nb);
-	if (err)
-		goto out;
-
-	/* We won't get a call on the boot CPU, so register immediately */
-	if (num_possible_cpus() > 1)
-		dummy_timer_setup();
-
-out:
-	cpu_notifier_register_done();
-	return err;
+	return cpuhp_setup_state(CPUHP_AP_DUMMY_TIMER_STARTING,
+				 "AP_DUMMY_TIMER_STARTING",
+				 dummy_timer_starting_cpu, NULL);
 }
 early_initcall(dummy_timer_register);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 672f690..1d19ed3 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -38,6 +38,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] clocksource/dummy_timer: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  00c1d17aab513d3b8117fb84644eba39434b33e4
Gitweb:     http://git.kernel.org/tip/00c1d17aab513d3b8117fb84644eba39434b33e4
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:41 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:24 +0200

clocksource/dummy_timer: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.130385842@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/dummy_timer.c | 36 ++++++------------------------------
 include/linux/cpuhotplug.h        |  1 +
 2 files changed, 7 insertions(+), 30 deletions(-)

diff --git a/drivers/clocksource/dummy_timer.c b/drivers/clocksource/dummy_timer.c
index 776b6c8..89f1c2e 100644
--- a/drivers/clocksource/dummy_timer.c
+++ b/drivers/clocksource/dummy_timer.c
@@ -16,10 +16,9 @@
 
 static DEFINE_PER_CPU(struct clock_event_device, dummy_timer_evt);
 
-static void dummy_timer_setup(void)
+static int dummy_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
-	struct clock_event_device *evt = raw_cpu_ptr(&dummy_timer_evt);
+	struct clock_event_device *evt = per_cpu_ptr(&dummy_timer_evt, cpu);
 
 	evt->name	= "dummy_timer";
 	evt->features	= CLOCK_EVT_FEAT_PERIODIC |
@@ -29,36 +28,13 @@ static void dummy_timer_setup(void)
 	evt->cpumask	= cpumask_of(cpu);
 
 	clockevents_register_device(evt);
+	return 0;
 }
 
-static int dummy_timer_cpu_notify(struct notifier_block *self,
-				      unsigned long action, void *hcpu)
-{
-	if ((action & ~CPU_TASKS_FROZEN) == CPU_STARTING)
-		dummy_timer_setup();
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block dummy_timer_cpu_nb = {
-	.notifier_call = dummy_timer_cpu_notify,
-};
-
 static int __init dummy_timer_register(void)
 {
-	int err = 0;
-
-	cpu_notifier_register_begin();
-	err = __register_cpu_notifier(&dummy_timer_cpu_nb);
-	if (err)
-		goto out;
-
-	/* We won't get a call on the boot CPU, so register immediately */
-	if (num_possible_cpus() > 1)
-		dummy_timer_setup();
-
-out:
-	cpu_notifier_register_done();
-	return err;
+	return cpuhp_setup_state(CPUHP_AP_DUMMY_TIMER_STARTING,
+				 "AP_DUMMY_TIMER_STARTING",
+				 dummy_timer_starting_cpu, NULL);
 }
 early_initcall(dummy_timer_register);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f0524cd..8103561 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -37,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 40/67] clocksource/metag: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0039-clocksource-metag-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2791 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-metag@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/metag_generic.c | 33 ++++++---------------------------
 include/linux/cpuhotplug.h          |  1 +
 2 files changed, 7 insertions(+), 27 deletions(-)

diff --git a/drivers/clocksource/metag_generic.c b/drivers/clocksource/metag_generic.c
index bcd5c0d..a80ab3e 100644
--- a/drivers/clocksource/metag_generic.c
+++ b/drivers/clocksource/metag_generic.c
@@ -90,7 +90,7 @@ unsigned long long sched_clock(void)
 	return ticks << HARDWARE_TO_NS_SHIFT;
 }
 
-static void arch_timer_setup(unsigned int cpu)
+static int arch_timer_starting_cpu(unsigned int cpu)
 {
 	unsigned int txdivtime;
 	struct clock_event_device *clk = &per_cpu(local_clockevent, cpu);
@@ -132,27 +132,9 @@ static void arch_timer_setup(unsigned int cpu)
 		val = core_reg_read(TXUCT_ID, TXTIMER_REGNUM, thread0);
 		__core_reg_set(TXTIMER, val);
 	}
+	return 0;
 }
 
-static int arch_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	int cpu = (long)hcpu;
-
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		arch_timer_setup(cpu);
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block arch_timer_cpu_nb = {
-	.notifier_call = arch_timer_cpu_notify,
-};
-
 int __init metag_generic_timer_init(void)
 {
 	/*
@@ -170,11 +152,8 @@ int __init metag_generic_timer_init(void)
 
 	setup_irq(tbisig_map(TBID_SIGNUM_TRT), &metag_timer_irq);
 
-	/* Configure timer on boot CPU */
-	arch_timer_setup(smp_processor_id());
-
-	/* Hook cpu boot to configure other CPU's timers */
-	register_cpu_notifier(&arch_timer_cpu_nb);
-
-	return 0;
+	/* Hook cpu boot to configure the CPU's timers */
+	return cpuhp_setup_state(CPUHP_AP_METAG_TIMER_STARTING,
+				 "AP_METAG_TIMER_STARTING",
+				 arch_timer_starting_cpu, NULL);
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 1d19ed3..2d20836 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -39,6 +39,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] clocksource/metag: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  31e8e5db8849e4c7084dd5bce7b7d96d2fb14b8b
Gitweb:     http://git.kernel.org/tip/31e8e5db8849e4c7084dd5bce7b7d96d2fb14b8b
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:42 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:25 +0200

clocksource/metag: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-metag@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.215137642@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/metag_generic.c | 33 ++++++---------------------------
 include/linux/cpuhotplug.h          |  1 +
 2 files changed, 7 insertions(+), 27 deletions(-)

diff --git a/drivers/clocksource/metag_generic.c b/drivers/clocksource/metag_generic.c
index bcd5c0d..a80ab3e 100644
--- a/drivers/clocksource/metag_generic.c
+++ b/drivers/clocksource/metag_generic.c
@@ -90,7 +90,7 @@ unsigned long long sched_clock(void)
 	return ticks << HARDWARE_TO_NS_SHIFT;
 }
 
-static void arch_timer_setup(unsigned int cpu)
+static int arch_timer_starting_cpu(unsigned int cpu)
 {
 	unsigned int txdivtime;
 	struct clock_event_device *clk = &per_cpu(local_clockevent, cpu);
@@ -132,27 +132,9 @@ static void arch_timer_setup(unsigned int cpu)
 		val = core_reg_read(TXUCT_ID, TXTIMER_REGNUM, thread0);
 		__core_reg_set(TXTIMER, val);
 	}
+	return 0;
 }
 
-static int arch_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	int cpu = (long)hcpu;
-
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		arch_timer_setup(cpu);
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block arch_timer_cpu_nb = {
-	.notifier_call = arch_timer_cpu_notify,
-};
-
 int __init metag_generic_timer_init(void)
 {
 	/*
@@ -170,11 +152,8 @@ int __init metag_generic_timer_init(void)
 
 	setup_irq(tbisig_map(TBID_SIGNUM_TRT), &metag_timer_irq);
 
-	/* Configure timer on boot CPU */
-	arch_timer_setup(smp_processor_id());
-
-	/* Hook cpu boot to configure other CPU's timers */
-	register_cpu_notifier(&arch_timer_cpu_nb);
-
-	return 0;
+	/* Hook cpu boot to configure the CPU's timers */
+	return cpuhp_setup_state(CPUHP_AP_METAG_TIMER_STARTING,
+				 "AP_METAG_TIMER_STARTING",
+				 arch_timer_starting_cpu, NULL);
 }
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 8103561..afba3b1 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -38,6 +38,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 41/67] clocksource/qcom-timer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0040-clocksource-qcom-timer-Convert-to-hotplug-state-mach.patch --]
[-- Type: text/plain, Size: 3292 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/qcom-timer.c | 41 +++++++++++-----------------------------
 include/linux/cpuhotplug.h       |  1 +
 2 files changed, 12 insertions(+), 30 deletions(-)

diff --git a/drivers/clocksource/qcom-timer.c b/drivers/clocksource/qcom-timer.c
index 6625763..3283cfa 100644
--- a/drivers/clocksource/qcom-timer.c
+++ b/drivers/clocksource/qcom-timer.c
@@ -105,9 +105,9 @@ static struct clocksource msm_clocksource = {
 static int msm_timer_irq;
 static int msm_timer_has_ppi;
 
-static int msm_local_timer_setup(struct clock_event_device *evt)
+static int msm_local_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *evt = per_cpu_ptr(msm_evt, cpu);
 	int err;
 
 	evt->irq = msm_timer_irq;
@@ -135,35 +135,15 @@ static int msm_local_timer_setup(struct clock_event_device *evt)
 	return 0;
 }
 
-static void msm_local_timer_stop(struct clock_event_device *evt)
+static int msm_local_timer_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(msm_evt, cpu);
+
 	evt->set_state_shutdown(evt);
 	disable_percpu_irq(evt->irq);
+	return 0;
 }
 
-static int msm_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		msm_local_timer_setup(this_cpu_ptr(msm_evt));
-		break;
-	case CPU_DYING:
-		msm_local_timer_stop(this_cpu_ptr(msm_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block msm_timer_cpu_nb = {
-	.notifier_call = msm_timer_cpu_notify,
-};
-
 static u64 notrace msm_sched_clock_read(void)
 {
 	return msm_clocksource.read(&msm_clocksource);
@@ -200,14 +180,15 @@ static int __init msm_timer_init(u32 dgt_hz, int sched_bits, int irq,
 	if (res) {
 		pr_err("request_percpu_irq failed\n");
 	} else {
-		res = register_cpu_notifier(&msm_timer_cpu_nb);
+		/* Install and invoke hotplug callbacks */
+		res = cpuhp_setup_state(CPUHP_AP_QCOM_TIMER_STARTING,
+					"AP_QCOM_TIMER_STARTING",
+					msm_local_timer_starting_cpu,
+					msm_local_timer_dying_cpu);
 		if (res) {
 			free_percpu_irq(irq, msm_evt);
 			goto err;
 		}
-
-		/* Immediately configure the timer on the boot CPU */
-		msm_local_timer_setup(raw_cpu_ptr(msm_evt));
 	}
 
 err:
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 2d20836..5121ec1 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -40,6 +40,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
+	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] clocksource/qcom-timer: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  b04041655d0cf84b77802cb0b4517e672df04f01
Gitweb:     http://git.kernel.org/tip/b04041655d0cf84b77802cb0b4517e672df04f01
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:43 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:25 +0200

clocksource/qcom-timer: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.295486558@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/qcom-timer.c | 41 +++++++++++-----------------------------
 include/linux/cpuhotplug.h       |  1 +
 2 files changed, 12 insertions(+), 30 deletions(-)

diff --git a/drivers/clocksource/qcom-timer.c b/drivers/clocksource/qcom-timer.c
index 6625763..3283cfa 100644
--- a/drivers/clocksource/qcom-timer.c
+++ b/drivers/clocksource/qcom-timer.c
@@ -105,9 +105,9 @@ static struct clocksource msm_clocksource = {
 static int msm_timer_irq;
 static int msm_timer_has_ppi;
 
-static int msm_local_timer_setup(struct clock_event_device *evt)
+static int msm_local_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *evt = per_cpu_ptr(msm_evt, cpu);
 	int err;
 
 	evt->irq = msm_timer_irq;
@@ -135,35 +135,15 @@ static int msm_local_timer_setup(struct clock_event_device *evt)
 	return 0;
 }
 
-static void msm_local_timer_stop(struct clock_event_device *evt)
+static int msm_local_timer_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(msm_evt, cpu);
+
 	evt->set_state_shutdown(evt);
 	disable_percpu_irq(evt->irq);
+	return 0;
 }
 
-static int msm_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		msm_local_timer_setup(this_cpu_ptr(msm_evt));
-		break;
-	case CPU_DYING:
-		msm_local_timer_stop(this_cpu_ptr(msm_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block msm_timer_cpu_nb = {
-	.notifier_call = msm_timer_cpu_notify,
-};
-
 static u64 notrace msm_sched_clock_read(void)
 {
 	return msm_clocksource.read(&msm_clocksource);
@@ -200,14 +180,15 @@ static int __init msm_timer_init(u32 dgt_hz, int sched_bits, int irq,
 	if (res) {
 		pr_err("request_percpu_irq failed\n");
 	} else {
-		res = register_cpu_notifier(&msm_timer_cpu_nb);
+		/* Install and invoke hotplug callbacks */
+		res = cpuhp_setup_state(CPUHP_AP_QCOM_TIMER_STARTING,
+					"AP_QCOM_TIMER_STARTING",
+					msm_local_timer_starting_cpu,
+					msm_local_timer_dying_cpu);
 		if (res) {
 			free_percpu_irq(irq, msm_evt);
 			goto err;
 		}
-
-		/* Immediately configure the timer on the boot CPU */
-		msm_local_timer_setup(raw_cpu_ptr(msm_evt));
 	}
 
 err:
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index afba3b1..d74d15a 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -39,6 +39,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
+	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 42/67] clocksource/mips-gic: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0041-clocksource-mips-gic-Convert-to-hotplug-state-machin.patch --]
[-- Type: text/plain, Size: 3270 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/mips-gic-timer.c | 38 ++++++++++++------------------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 14 insertions(+), 25 deletions(-)

diff --git a/drivers/clocksource/mips-gic-timer.c b/drivers/clocksource/mips-gic-timer.c
index 1572c7a..d91e872 100644
--- a/drivers/clocksource/mips-gic-timer.c
+++ b/drivers/clocksource/mips-gic-timer.c
@@ -49,10 +49,9 @@ struct irqaction gic_compare_irqaction = {
 	.name = "timer",
 };
 
-static void gic_clockevent_cpu_init(struct clock_event_device *cd)
+static void gic_clockevent_cpu_init(unsigned int cpu,
+				    struct clock_event_device *cd)
 {
-	unsigned int cpu = smp_processor_id();
-
 	cd->name		= "MIPS GIC";
 	cd->features		= CLOCK_EVT_FEAT_ONESHOT |
 				  CLOCK_EVT_FEAT_C3STOP;
@@ -79,19 +78,10 @@ static void gic_update_frequency(void *data)
 	clockevents_update_freq(this_cpu_ptr(&gic_clockevent_device), rate);
 }
 
-static int gic_cpu_notifier(struct notifier_block *nb, unsigned long action,
-				void *data)
+static int gic_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		gic_clockevent_cpu_init(this_cpu_ptr(&gic_clockevent_device));
-		break;
-	case CPU_DYING:
-		gic_clockevent_cpu_exit(this_cpu_ptr(&gic_clockevent_device));
-		break;
-	}
-
-	return NOTIFY_OK;
+	gic_clockevent_cpu_init(cpu, this_cpu_ptr(&gic_clockevent_device));
+	return 0;
 }
 
 static int gic_clk_notifier(struct notifier_block *nb, unsigned long action,
@@ -105,10 +95,11 @@ static int gic_clk_notifier(struct notifier_block *nb, unsigned long action,
 	return NOTIFY_OK;
 }
 
-
-static struct notifier_block gic_cpu_nb = {
-	.notifier_call = gic_cpu_notifier,
-};
+static int gic_dying_cpu(unsigned int cpu)
+{
+	gic_clockevent_cpu_exit(this_cpu_ptr(&gic_clockevent_device));
+	return 0;
+}
 
 static struct notifier_block gic_clk_nb = {
 	.notifier_call = gic_clk_notifier,
@@ -125,12 +116,9 @@ static int gic_clockevent_init(void)
 	if (ret < 0)
 		return ret;
 
-	ret = register_cpu_notifier(&gic_cpu_nb);
-	if (ret < 0)
-		pr_warn("GIC: Unable to register CPU notifier\n");
-
-	gic_clockevent_cpu_init(this_cpu_ptr(&gic_clockevent_device));
-
+	cpuhp_setup_state(CPUHP_AP_MIPS_GIC_TIMER_STARTING,
+			  "AP_MIPS_GIC_TIMER_STARTING", gic_starting_cpu,
+			  gic_dying_cpu);
 	return 0;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 5121ec1..c052b04 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -41,6 +41,7 @@ enum cpuhp_state {
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
+	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] clocksource/mips-gic: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  2dab90932fd08f101a45acc0495ccd77b869da5d
Gitweb:     http://git.kernel.org/tip/2dab90932fd08f101a45acc0495ccd77b869da5d
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:44 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:26 +0200

clocksource/mips-gic: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.380737946@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/mips-gic-timer.c | 38 ++++++++++++------------------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 14 insertions(+), 25 deletions(-)

diff --git a/drivers/clocksource/mips-gic-timer.c b/drivers/clocksource/mips-gic-timer.c
index 1572c7a..d91e872 100644
--- a/drivers/clocksource/mips-gic-timer.c
+++ b/drivers/clocksource/mips-gic-timer.c
@@ -49,10 +49,9 @@ struct irqaction gic_compare_irqaction = {
 	.name = "timer",
 };
 
-static void gic_clockevent_cpu_init(struct clock_event_device *cd)
+static void gic_clockevent_cpu_init(unsigned int cpu,
+				    struct clock_event_device *cd)
 {
-	unsigned int cpu = smp_processor_id();
-
 	cd->name		= "MIPS GIC";
 	cd->features		= CLOCK_EVT_FEAT_ONESHOT |
 				  CLOCK_EVT_FEAT_C3STOP;
@@ -79,19 +78,10 @@ static void gic_update_frequency(void *data)
 	clockevents_update_freq(this_cpu_ptr(&gic_clockevent_device), rate);
 }
 
-static int gic_cpu_notifier(struct notifier_block *nb, unsigned long action,
-				void *data)
+static int gic_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		gic_clockevent_cpu_init(this_cpu_ptr(&gic_clockevent_device));
-		break;
-	case CPU_DYING:
-		gic_clockevent_cpu_exit(this_cpu_ptr(&gic_clockevent_device));
-		break;
-	}
-
-	return NOTIFY_OK;
+	gic_clockevent_cpu_init(cpu, this_cpu_ptr(&gic_clockevent_device));
+	return 0;
 }
 
 static int gic_clk_notifier(struct notifier_block *nb, unsigned long action,
@@ -105,10 +95,11 @@ static int gic_clk_notifier(struct notifier_block *nb, unsigned long action,
 	return NOTIFY_OK;
 }
 
-
-static struct notifier_block gic_cpu_nb = {
-	.notifier_call = gic_cpu_notifier,
-};
+static int gic_dying_cpu(unsigned int cpu)
+{
+	gic_clockevent_cpu_exit(this_cpu_ptr(&gic_clockevent_device));
+	return 0;
+}
 
 static struct notifier_block gic_clk_nb = {
 	.notifier_call = gic_clk_notifier,
@@ -125,12 +116,9 @@ static int gic_clockevent_init(void)
 	if (ret < 0)
 		return ret;
 
-	ret = register_cpu_notifier(&gic_cpu_nb);
-	if (ret < 0)
-		pr_warn("GIC: Unable to register CPU notifier\n");
-
-	gic_clockevent_cpu_init(this_cpu_ptr(&gic_clockevent_device));
-
+	cpuhp_setup_state(CPUHP_AP_MIPS_GIC_TIMER_STARTING,
+			  "AP_MIPS_GIC_TIMER_STARTING", gic_starting_cpu,
+			  gic_dying_cpu);
 	return 0;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d74d15a..456e3d9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -40,6 +40,7 @@ enum cpuhp_state {
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
+	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0042-leds-trigger-cpu-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2713 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

This is a straightforward conversion. We place this callback last
in the list so that the LED illuminates only after a successful
bring up sequence.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Jacek Anaszewski <j.anaszewski@samsung.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Richard Purdie <rpurdie@rpsys.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-leds@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/leds/trigger/ledtrig-cpu.c | 32 +++++++++++++++-----------------
 include/linux/cpuhotplug.h         |  1 +
 2 files changed, 16 insertions(+), 17 deletions(-)

diff --git a/drivers/leds/trigger/ledtrig-cpu.c b/drivers/leds/trigger/ledtrig-cpu.c
index 938467f..4a6a182 100644
--- a/drivers/leds/trigger/ledtrig-cpu.c
+++ b/drivers/leds/trigger/ledtrig-cpu.c
@@ -92,25 +92,17 @@ static struct syscore_ops ledtrig_cpu_syscore_ops = {
 	.resume		= ledtrig_cpu_syscore_resume,
 };
 
-static int ledtrig_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
+static int ledtrig_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		ledtrig_cpu(CPU_LED_START);
-		break;
-	case CPU_DYING:
-		ledtrig_cpu(CPU_LED_STOP);
-		break;
-	}
-
-	return NOTIFY_OK;
+	ledtrig_cpu(CPU_LED_START);
+	return 0;
 }
 
-
-static struct notifier_block ledtrig_cpu_nb = {
-	.notifier_call = ledtrig_cpu_notify,
-};
+static int ledtrig_dying_cpu(unsigned int cpu)
+{
+	ledtrig_cpu(CPU_LED_STOP);
+	return 0;
+}
 
 static int __init ledtrig_cpu_init(void)
 {
@@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
 	}
 
 	register_syscore_ops(&ledtrig_cpu_syscore_ops);
-	register_cpu_notifier(&ledtrig_cpu_nb);
+
+	/*
+	 * FIXME: Why needs this to happen in the interrupt disabled
+	 * low level bringup phase of a cpu?
+	 */
+	cpuhp_setup_state(CPUHP_AP_LEDTRIG_STARTING, "AP_LEDTRIG_STARTING",
+			  ledtrig_starting_cpu, ledtrig_dying_cpu);
 
 	pr_info("ledtrig-cpu: registered to indicate activity on CPUs\n");
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index c052b04..ef7bfa6 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -43,6 +43,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Jacek Anaszewski]

Hi Anna,

On 07/13/2016 07:16 PM, Anna-Maria Gleixner wrote:

 > -------- Original Message --------
 > Subject: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state 
machine
 > Date: Wed, 13 Jul 2016 17:16:45 +0000
 > From: Anna-Maria Gleixner <anna-maria@linutronix.de>
 > To: LKML <linux-kernel@vger.kernel.org>
 > CC: Peter Zijlstra <peterz@infradead.org>, Ingo Molnar 
<mingo@kernel.org>, Thomas Gleixner <tglx@linutronix.de>, 
rt@linutronix.de, Richard Cochran <rcochran@linutronix.de>, Sebastian 
Andrzej Siewior <bigeasy@linutronix.de>, Jacek Anaszewski 
<j.anaszewski@samsung.com>, Linus Torvalds 
<torvalds@linux-foundation.org>, Linus Walleij 
<linus.walleij@linaro.org>, Paul Gortmaker 
<paul.gortmaker@windriver.com>, Richard Purdie <rpurdie@rpsys.net>, 
linux-leds@vger.kernel.org, Anna-Maria Gleixner <anna-maria@linutronix.de>
 >
 > From: Richard Cochran <rcochran@linutronix.de>
 >
 > This is a straightforward conversion. We place this callback last
 > in the list so that the LED illuminates only after a successful
 > bring up sequence.
 >
 > Signed-off-by: Richard Cochran <rcochran@linutronix.de>
 > Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
 > Cc: Jacek Anaszewski <j.anaszewski@samsung.com>
 > Cc: Linus Torvalds <torvalds@linux-foundation.org>
 > Cc: Linus Walleij <linus.walleij@linaro.org>
 > Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
 > Cc: Peter Zijlstra <peterz@infradead.org>
 > Cc: Richard Purdie <rpurdie@rpsys.net>
 > Cc: Thomas Gleixner <tglx@linutronix.de>
 > Cc: linux-leds@vger.kernel.org
 > Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
 > ---
 >  drivers/leds/trigger/ledtrig-cpu.c | 32 +++++++++++++++-----------------
 >  include/linux/cpuhotplug.h         |  1 +
 >  2 files changed, 16 insertions(+), 17 deletions(-)
 >
 > diff --git a/drivers/leds/trigger/ledtrig-cpu.c 
b/drivers/leds/trigger/ledtrig-cpu.c
 > index 938467f..4a6a182 100644
 > --- a/drivers/leds/trigger/ledtrig-cpu.c
 > +++ b/drivers/leds/trigger/ledtrig-cpu.c
 > @@ -92,25 +92,17 @@ static struct syscore_ops ledtrig_cpu_syscore_ops = {
 >       .resume         = ledtrig_cpu_syscore_resume,
 >  };
 >
 > -static int ledtrig_cpu_notify(struct notifier_block *self,
 > -                                        unsigned long action, void 
*hcpu)
 > +static int ledtrig_starting_cpu(unsigned int cpu)
 >  {
 > -     switch (action & ~CPU_TASKS_FROZEN) {
 > -     case CPU_STARTING:
 > -             ledtrig_cpu(CPU_LED_START);
 > -             break;
 > -             break;
 > -     }
 > -
 > -     return NOTIFY_OK;
 > +     ledtrig_cpu(CPU_LED_START);
 > +     return 0;
 >  }
 >
 > -
 > -static struct notifier_block ledtrig_cpu_nb = {
 > -     .notifier_call = ledtrig_cpu_notify,
 > -};
 > +static int ledtrig_dying_cpu(unsigned int cpu)
 > +{
 > +     ledtrig_cpu(CPU_LED_STOP);
 > +     return 0;
 > +}
 >
 >  static int __init ledtrig_cpu_init(void)
 >  {
 > @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
 >       }
 >
 >       register_syscore_ops(&ledtrig_cpu_syscore_ops);
 > -     register_cpu_notifier(&ledtrig_cpu_nb);
 > +
 > +     /*
 > +      * FIXME: Why needs this to happen in the interrupt disabled
 > +      * low level bringup phase of a cpu?
 > +      */

Why wasn't it possible to clarify the issue before the
submission?

 > +     cpuhp_setup_state(CPUHP_AP_LEDTRIG_STARTING, "AP_LEDTRIG_STARTING",
 > +                       ledtrig_starting_cpu, ledtrig_dying_cpu);
 >
 >       pr_info("ledtrig-cpu: registered to indicate activity on CPUs\n");
 >
 > diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
 > index c052b04..ef7bfa6 100644
 > --- a/include/linux/cpuhotplug.h
 > +++ b/include/linux/cpuhotplug.h
 > @@ -43,6 +43,7 @@ enum cpuhp_state {
 >       CPUHP_AP_QCOM_TIMER_STARTING,
 >       CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 >       CPUHP_AP_KVM_STARTING,
 > +     CPUHP_AP_LEDTRIG_STARTING,
 >       CPUHP_AP_NOTIFY_STARTING,
 >       CPUHP_AP_ONLINE,
 >       CPUHP_TEARDOWN_CPU,
 > --


-- 
Best regards,
Jacek Anaszewski

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Ingo Molnar]

* Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> > @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
> >       }
> >
> >       register_syscore_ops(&ledtrig_cpu_syscore_ops);
> > -     register_cpu_notifier(&ledtrig_cpu_nb);
> > +
> > +     /*
> > +      * FIXME: Why needs this to happen in the interrupt disabled
> > +      * low level bringup phase of a cpu?
> > +      */
> 
> Why wasn't it possible to clarify the issue before the
> submission?

Nothing has been applied yet (this submission is part of the clarification 
process), are you fine with this conversion as-is, with a separate patch doing any 
followup fixes - or would you prefer another approach?

This CPU hotplug series tries to do straightforward conversions to the CPU hotplug 
state machine without changing behavior - while pointing out any problems that 
were found along the way.

I agree that the addition of this FIXME should probably have been pointed out in 
the changelog as well, but this patch was submitted once already with no feedback 
received - and that's the historic pattern with such types of series: only a low 
percentage of all driver maintainers replies so we have to be proactive to a 
certain degree to have a chance of improving core kernel infrastructure. It's 
better to add FIXMEs with open questions that to introduce subtle problems.

Thanks,

	Ingo

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Jacek Anaszewski]

On 07/14/2016 09:47 AM, Ingo Molnar wrote:
>
> * Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>>> @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
>>>        }
>>>
>>>        register_syscore_ops(&ledtrig_cpu_syscore_ops);
>>> -     register_cpu_notifier(&ledtrig_cpu_nb);
>>> +
>>> +     /*
>>> +      * FIXME: Why needs this to happen in the interrupt disabled
>>> +      * low level bringup phase of a cpu?
>>> +      */
>>
>> Why wasn't it possible to clarify the issue before the
>> submission?
>
> Nothing has been applied yet (this submission is part of the clarification
> process), are you fine with this conversion as-is, with a separate patch doing any
> followup fixes - or would you prefer another approach?
>
> This CPU hotplug series tries to do straightforward conversions to the CPU hotplug
> state machine without changing behavior - while pointing out any problems that
> were found along the way.
>
> I agree that the addition of this FIXME should probably have been pointed out in
> the changelog as well, but this patch was submitted once already with no feedback
> received - and that's the historic pattern with such types of series: only a low
> percentage of all driver maintainers replies so we have to be proactive to a
> certain degree to have a chance of improving core kernel infrastructure. It's
> better to add FIXMEs with open questions that to introduce subtle problems.

Thanks for the explanation. I'm OK with that approach.

Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>

-- 
Best regards,
Jacek Anaszewski

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Peter Zijlstra]

On Thu, Jul 14, 2016 at 10:10:55AM +0200, Jacek Anaszewski wrote:
> On 07/14/2016 09:47 AM, Ingo Molnar wrote:
> >
> >* Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >
> >>>@@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
> >>>       }
> >>>
> >>>       register_syscore_ops(&ledtrig_cpu_syscore_ops);
> >>>-     register_cpu_notifier(&ledtrig_cpu_nb);
> >>>+
> >>>+     /*
> >>>+      * FIXME: Why needs this to happen in the interrupt disabled
> >>>+      * low level bringup phase of a cpu?
> >>>+      */

> Thanks for the explanation. I'm OK with that approach.

Since we have your attention, could you perhaps attempt to answer the
question so we can fix the fixme?

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Jacek Anaszewski]

On 07/14/2016 11:41 AM, Peter Zijlstra wrote:
> On Thu, Jul 14, 2016 at 10:10:55AM +0200, Jacek Anaszewski wrote:
>> On 07/14/2016 09:47 AM, Ingo Molnar wrote:
>>>
>>> * Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>
>>>>> @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
>>>>>        }
>>>>>
>>>>>        register_syscore_ops(&ledtrig_cpu_syscore_ops);
>>>>> -     register_cpu_notifier(&ledtrig_cpu_nb);
>>>>> +
>>>>> +     /*
>>>>> +      * FIXME: Why needs this to happen in the interrupt disabled
>>>>> +      * low level bringup phase of a cpu?
>>>>> +      */
>
>> Thanks for the explanation. I'm OK with that approach.
>
> Since we have your attention, could you perhaps attempt to answer the
> question so we can fix the fixme?
>

This is the way how all led triggers are being registered.

Once a trigger module is loaded it needs to be registered in
the LED Trigger core.

-- 
Best regards,
Jacek Anaszewski

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Thomas Gleixner]

On Thu, 14 Jul 2016, Jacek Anaszewski wrote:
> On 07/14/2016 11:41 AM, Peter Zijlstra wrote:
> > On Thu, Jul 14, 2016 at 10:10:55AM +0200, Jacek Anaszewski wrote:
> > > On 07/14/2016 09:47 AM, Ingo Molnar wrote:
> > > > 
> > > > * Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> > > > 
> > > > > > @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
> > > > > >        }
> > > > > > 
> > > > > >        register_syscore_ops(&ledtrig_cpu_syscore_ops);
> > > > > > -     register_cpu_notifier(&ledtrig_cpu_nb);
> > > > > > +
> > > > > > +     /*
> > > > > > +      * FIXME: Why needs this to happen in the interrupt disabled
> > > > > > +      * low level bringup phase of a cpu?
> > > > > > +      */
> > 
> > > Thanks for the explanation. I'm OK with that approach.
> > 
> > Since we have your attention, could you perhaps attempt to answer the
> > question so we can fix the fixme?
> > 
> 
> This is the way how all led triggers are being registered.
> 
> Once a trigger module is loaded it needs to be registered in
> the LED Trigger core.

That does not explain WHY this needs to happen in the low level bringup phase
of the CPU with interrupts disabled and can't be done from the normal ONLINE
callbacks in thread context.

Thanks,

	tglx

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Jacek Anaszewski]

On 07/14/2016 01:33 PM, Thomas Gleixner wrote:
> On Thu, 14 Jul 2016, Jacek Anaszewski wrote:
>> On 07/14/2016 11:41 AM, Peter Zijlstra wrote:
>>> On Thu, Jul 14, 2016 at 10:10:55AM +0200, Jacek Anaszewski wrote:
>>>> On 07/14/2016 09:47 AM, Ingo Molnar wrote:
>>>>>
>>>>> * Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>>>
>>>>>>> @@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
>>>>>>>         }
>>>>>>>
>>>>>>>         register_syscore_ops(&ledtrig_cpu_syscore_ops);
>>>>>>> -     register_cpu_notifier(&ledtrig_cpu_nb);
>>>>>>> +
>>>>>>> +     /*
>>>>>>> +      * FIXME: Why needs this to happen in the interrupt disabled
>>>>>>> +      * low level bringup phase of a cpu?
>>>>>>> +      */
>>>
>>>> Thanks for the explanation. I'm OK with that approach.
>>>
>>> Since we have your attention, could you perhaps attempt to answer the
>>> question so we can fix the fixme?
>>>
>>
>> This is the way how all led triggers are being registered.
>>
>> Once a trigger module is loaded it needs to be registered in
>> the LED Trigger core.
>
> That does not explain WHY this needs to happen in the low level bringup phase
> of the CPU with interrupts disabled and can't be done from the normal ONLINE
> callbacks in thread context.

It was before my time in kernel, so I can only suppose that it was
the easiest way. Does it introduce some problems?

-- 
Best regards,
Jacek Anaszewski

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

* Jacek Anaszewski | 2016-07-14 13:55:26 [+0200]:

>On 07/14/2016 01:33 PM, Thomas Gleixner wrote:
>>That does not explain WHY this needs to happen in the low level bringup phase
>>of the CPU with interrupts disabled and can't be done from the normal ONLINE
>>callbacks in thread context.
>
>It was before my time in kernel, so I can only suppose that it was
>the easiest way. Does it introduce some problems?

It was introduced by Pawel Moll in fba14ae8e924 ("ledtrig-cpu: Handle
CPU hot(un)plugging"). It does not introduce any problems but those
bring up / bring down levels are usually used by the core code. This
does not look like it needs to be done _that_ early or late during the
removal / addition of a CPU to the system.

Which means this looks like it could be moved to CPU_ONLINE /
CPU_DOWN_PREPARE which in turns means we could use dynamic IDs instead
of those hardcoded ones since the LED subsystem does not depend on other
components.

Sebastian

Re: [patch V2 43/67] leds/trigger/cpu: Convert to hotplug state machine

[Jacek Anaszewski]

Hi Sebastian,

On 07/15/2016 04:10 PM, Sebastian Andrzej Siewior wrote:
> * Jacek Anaszewski | 2016-07-14 13:55:26 [+0200]:
>
>> On 07/14/2016 01:33 PM, Thomas Gleixner wrote:
>>> That does not explain WHY this needs to happen in the low level bringup phase
>>> of the CPU with interrupts disabled and can't be done from the normal ONLINE
>>> callbacks in thread context.
>>
>> It was before my time in kernel, so I can only suppose that it was
>> the easiest way. Does it introduce some problems?
>
> It was introduced by Pawel Moll in fba14ae8e924 ("ledtrig-cpu: Handle
> CPU hot(un)plugging"). It does not introduce any problems but those
> bring up / bring down levels are usually used by the core code. This
> does not look like it needs to be done _that_ early or late during the
> removal / addition of a CPU to the system.
>
> Which means this looks like it could be moved to CPU_ONLINE /
> CPU_DOWN_PREPARE which in turns means we could use dynamic IDs instead
> of those hardcoded ones since the LED subsystem does not depend on other
> components.

 From the LED subsystem perspective I don't see any particular reason
for which it couldn't be accomplished from the ONLINE callbacks.

-- 
Best regards,
Jacek Anaszewski

[tip:smp/hotplug] leds/trigger/cpu: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  911a359de96d5e9d24f50c74a48c6f3cf061539a
Gitweb:     http://git.kernel.org/tip/911a359de96d5e9d24f50c74a48c6f3cf061539a
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:45 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:26 +0200

leds/trigger/cpu: Convert to hotplug state machine

This is a straightforward conversion. We place this callback last
in the list so that the LED illuminates only after a successful
bring up sequence.

( NOTE: The patch adds a FIXME question about the callback used,
        this question should probably be revisited later on.)

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Richard Purdie <rpurdie@rpsys.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-leds@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.465496902@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/leds/trigger/ledtrig-cpu.c | 32 +++++++++++++++-----------------
 include/linux/cpuhotplug.h         |  1 +
 2 files changed, 16 insertions(+), 17 deletions(-)

diff --git a/drivers/leds/trigger/ledtrig-cpu.c b/drivers/leds/trigger/ledtrig-cpu.c
index 938467f..4a6a182 100644
--- a/drivers/leds/trigger/ledtrig-cpu.c
+++ b/drivers/leds/trigger/ledtrig-cpu.c
@@ -92,25 +92,17 @@ static struct syscore_ops ledtrig_cpu_syscore_ops = {
 	.resume		= ledtrig_cpu_syscore_resume,
 };
 
-static int ledtrig_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
+static int ledtrig_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		ledtrig_cpu(CPU_LED_START);
-		break;
-	case CPU_DYING:
-		ledtrig_cpu(CPU_LED_STOP);
-		break;
-	}
-
-	return NOTIFY_OK;
+	ledtrig_cpu(CPU_LED_START);
+	return 0;
 }
 
-
-static struct notifier_block ledtrig_cpu_nb = {
-	.notifier_call = ledtrig_cpu_notify,
-};
+static int ledtrig_dying_cpu(unsigned int cpu)
+{
+	ledtrig_cpu(CPU_LED_STOP);
+	return 0;
+}
 
 static int __init ledtrig_cpu_init(void)
 {
@@ -133,7 +125,13 @@ static int __init ledtrig_cpu_init(void)
 	}
 
 	register_syscore_ops(&ledtrig_cpu_syscore_ops);
-	register_cpu_notifier(&ledtrig_cpu_nb);
+
+	/*
+	 * FIXME: Why needs this to happen in the interrupt disabled
+	 * low level bringup phase of a cpu?
+	 */
+	cpuhp_setup_state(CPUHP_AP_LEDTRIG_STARTING, "AP_LEDTRIG_STARTING",
+			  ledtrig_starting_cpu, ledtrig_dying_cpu);
 
 	pr_info("ledtrig-cpu: registered to indicate activity on CPUs\n");
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 456e3d9..24f56e2 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -42,6 +42,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 44/67] arm/kvm/vgic: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0043-arm-kvm-vgic-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3027 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.
The VGIC callback is run after KVM's main callback since it reflects the
makefile order.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/arm/vgic.c        | 39 ++++++++-------------------------------
 2 files changed, 9 insertions(+), 31 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ef7bfa6..80c15ce 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -43,6 +43,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
diff --git a/virt/kvm/arm/vgic.c b/virt/kvm/arm/vgic.c
index c3bfbb9..67cb5e9 100644
--- a/virt/kvm/arm/vgic.c
+++ b/virt/kvm/arm/vgic.c
@@ -2326,32 +2326,18 @@ int vgic_has_attr_regs(const struct vgic_io_range *ranges, phys_addr_t offset)
 		return -ENXIO;
 }
 
-static void vgic_init_maintenance_interrupt(void *info)
+static int vgic_starting_cpu(unsigned int cpu)
 {
 	enable_percpu_irq(vgic->maint_irq, 0);
+	return 0;
 }
 
-static int vgic_cpu_notify(struct notifier_block *self,
-			   unsigned long action, void *cpu)
+static int vgic_dying_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		vgic_init_maintenance_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(vgic->maint_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	disable_percpu_irq(vgic->maint_irq);
+	return 0;
 }
 
-static struct notifier_block vgic_cpu_nb = {
-	.notifier_call = vgic_cpu_notify,
-};
-
 static int kvm_vgic_probe(void)
 {
 	const struct gic_kvm_info *gic_kvm_info;
@@ -2392,19 +2378,10 @@ int kvm_vgic_hyp_init(void)
 		return ret;
 	}
 
-	ret = __register_cpu_notifier(&vgic_cpu_nb);
-	if (ret) {
-		kvm_err("Cannot register vgic CPU notifier\n");
-		goto out_free_irq;
-	}
-
-	on_each_cpu(vgic_init_maintenance_interrupt, NULL, 1);
-
+	cpuhp_setup_state(CPUHP_AP_KVM_ARM_VGIC_STARTING,
+			  "AP_KVM_ARM_VGIC_STARTING", vgic_starting_cpu,
+			  vgic_dying_cpu);
 	return 0;
-
-out_free_irq:
-	free_percpu_irq(vgic->maint_irq, kvm_get_running_vcpus());
-	return ret;
 }
 
 int kvm_irq_map_gsi(struct kvm *kvm,
-- 
2.8.1

[tip:smp/hotplug] arm/kvm/vgic: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  42ec50b5f257b19bd5653c40933b0803d5c4179a
Gitweb:     http://git.kernel.org/tip/42ec50b5f257b19bd5653c40933b0803d5c4179a
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:46 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:26 +0200

arm/kvm/vgic: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.
The VGIC callback is run after KVM's main callback since it reflects the
makefile order.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.546953286@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/arm/vgic.c        | 39 ++++++++-------------------------------
 2 files changed, 9 insertions(+), 31 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 24f56e2..ff37cac 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -42,6 +42,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
diff --git a/virt/kvm/arm/vgic.c b/virt/kvm/arm/vgic.c
index c3bfbb9..67cb5e9 100644
--- a/virt/kvm/arm/vgic.c
+++ b/virt/kvm/arm/vgic.c
@@ -2326,32 +2326,18 @@ int vgic_has_attr_regs(const struct vgic_io_range *ranges, phys_addr_t offset)
 		return -ENXIO;
 }
 
-static void vgic_init_maintenance_interrupt(void *info)
+static int vgic_starting_cpu(unsigned int cpu)
 {
 	enable_percpu_irq(vgic->maint_irq, 0);
+	return 0;
 }
 
-static int vgic_cpu_notify(struct notifier_block *self,
-			   unsigned long action, void *cpu)
+static int vgic_dying_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		vgic_init_maintenance_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(vgic->maint_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	disable_percpu_irq(vgic->maint_irq);
+	return 0;
 }
 
-static struct notifier_block vgic_cpu_nb = {
-	.notifier_call = vgic_cpu_notify,
-};
-
 static int kvm_vgic_probe(void)
 {
 	const struct gic_kvm_info *gic_kvm_info;
@@ -2392,19 +2378,10 @@ int kvm_vgic_hyp_init(void)
 		return ret;
 	}
 
-	ret = __register_cpu_notifier(&vgic_cpu_nb);
-	if (ret) {
-		kvm_err("Cannot register vgic CPU notifier\n");
-		goto out_free_irq;
-	}
-
-	on_each_cpu(vgic_init_maintenance_interrupt, NULL, 1);
-
+	cpuhp_setup_state(CPUHP_AP_KVM_ARM_VGIC_STARTING,
+			  "AP_KVM_ARM_VGIC_STARTING", vgic_starting_cpu,
+			  vgic_dying_cpu);
 	return 0;
-
-out_free_irq:
-	free_percpu_irq(vgic->maint_irq, kvm_get_running_vcpus());
-	return ret;
 }
 
 int kvm_irq_map_gsi(struct kvm *kvm,

[patch V2 45/67] arm/kvm/arch_timer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0044-arm-kvm-arch_timer-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2990 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/arm/arch_timer.c  | 35 +++++++++++------------------------
 2 files changed, 12 insertions(+), 24 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 80c15ce..0a36e61 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -44,6 +44,7 @@ enum cpuhp_state {
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
+	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c
index e2d5b6f..4fde8c7 100644
--- a/virt/kvm/arm/arch_timer.c
+++ b/virt/kvm/arm/arch_timer.c
@@ -405,26 +405,17 @@ u64 kvm_arm_timer_get_reg(struct kvm_vcpu *vcpu, u64 regid)
 	return (u64)-1;
 }
 
-static int kvm_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *cpu)
+static int kvm_timer_starting_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		kvm_timer_init_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(host_vtimer_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	kvm_timer_init_interrupt(NULL);
+	return 0;
 }
 
-static struct notifier_block kvm_timer_cpu_nb = {
-	.notifier_call = kvm_timer_cpu_notify,
-};
+static int kvm_timer_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(host_vtimer_irq);
+	return 0;
+}
 
 int kvm_timer_hyp_init(void)
 {
@@ -449,12 +440,6 @@ int kvm_timer_hyp_init(void)
 		goto out;
 	}
 
-	err = __register_cpu_notifier(&kvm_timer_cpu_nb);
-	if (err) {
-		kvm_err("Cannot register timer CPU notifier\n");
-		goto out_free;
-	}
-
 	wqueue = create_singlethread_workqueue("kvm_arch_timer");
 	if (!wqueue) {
 		err = -ENOMEM;
@@ -462,8 +447,10 @@ int kvm_timer_hyp_init(void)
 	}
 
 	kvm_info("virtual timer IRQ%d\n", host_vtimer_irq);
-	on_each_cpu(kvm_timer_init_interrupt, NULL, 1);
 
+	cpuhp_setup_state(CPUHP_AP_KVM_ARM_TIMER_STARTING,
+			  "AP_KVM_ARM_TIMER_STARTING", kvm_timer_starting_cpu,
+			  kvm_timer_dying_cpu);
 	goto out;
 out_free:
 	free_percpu_irq(host_vtimer_irq, kvm_get_running_vcpus());
-- 
2.8.1

[tip:smp/hotplug] arm/kvm/arch_timer: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  b3c9950a5c75983c0a2fcc93760f53ba52da3c85
Gitweb:     http://git.kernel.org/tip/b3c9950a5c75983c0a2fcc93760f53ba52da3c85
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:47 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:27 +0200

arm/kvm/arch_timer: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.634155707@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  1 +
 virt/kvm/arm/arch_timer.c  | 35 +++++++++++------------------------
 2 files changed, 12 insertions(+), 24 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ff37cac..88986c9 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -43,6 +43,7 @@ enum cpuhp_state {
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
+	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c
index e2d5b6f..4fde8c7 100644
--- a/virt/kvm/arm/arch_timer.c
+++ b/virt/kvm/arm/arch_timer.c
@@ -405,26 +405,17 @@ u64 kvm_arm_timer_get_reg(struct kvm_vcpu *vcpu, u64 regid)
 	return (u64)-1;
 }
 
-static int kvm_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *cpu)
+static int kvm_timer_starting_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		kvm_timer_init_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(host_vtimer_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	kvm_timer_init_interrupt(NULL);
+	return 0;
 }
 
-static struct notifier_block kvm_timer_cpu_nb = {
-	.notifier_call = kvm_timer_cpu_notify,
-};
+static int kvm_timer_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(host_vtimer_irq);
+	return 0;
+}
 
 int kvm_timer_hyp_init(void)
 {
@@ -449,12 +440,6 @@ int kvm_timer_hyp_init(void)
 		goto out;
 	}
 
-	err = __register_cpu_notifier(&kvm_timer_cpu_nb);
-	if (err) {
-		kvm_err("Cannot register timer CPU notifier\n");
-		goto out_free;
-	}
-
 	wqueue = create_singlethread_workqueue("kvm_arch_timer");
 	if (!wqueue) {
 		err = -ENOMEM;
@@ -462,8 +447,10 @@ int kvm_timer_hyp_init(void)
 	}
 
 	kvm_info("virtual timer IRQ%d\n", host_vtimer_irq);
-	on_each_cpu(kvm_timer_init_interrupt, NULL, 1);
 
+	cpuhp_setup_state(CPUHP_AP_KVM_ARM_TIMER_STARTING,
+			  "AP_KVM_ARM_TIMER_STARTING", kvm_timer_starting_cpu,
+			  kvm_timer_dying_cpu);
 	goto out;
 out_free:
 	free_percpu_irq(host_vtimer_irq, kvm_get_running_vcpus());

[patch V2 46/67] metag/perf: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0045-metag-perf-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2890 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-metag@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/metag/kernel/perf/perf_event.c | 26 +++++++-------------------
 include/linux/cpuhotplug.h          |  1 +
 2 files changed, 8 insertions(+), 19 deletions(-)

diff --git a/arch/metag/kernel/perf/perf_event.c b/arch/metag/kernel/perf/perf_event.c
index 33a365f..052cba2 100644
--- a/arch/metag/kernel/perf/perf_event.c
+++ b/arch/metag/kernel/perf/perf_event.c
@@ -806,25 +806,16 @@ static struct metag_pmu _metag_pmu = {
 };
 
 /* PMU CPU hotplug notifier */
-static int metag_pmu_cpu_notify(struct notifier_block *b, unsigned long action,
-				void *hcpu)
+static int metag_pmu_starting_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned int)hcpu;
 	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
 
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
-
 	memset(cpuc, 0, sizeof(struct cpu_hw_events));
 	raw_spin_lock_init(&cpuc->pmu_lock);
 
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block metag_pmu_notifier = {
-	.notifier_call = metag_pmu_cpu_notify,
-};
-
 /* PMU Initialisation */
 static int __init init_hw_perf_events(void)
 {
@@ -876,16 +867,13 @@ static int __init init_hw_perf_events(void)
 	metag_out32(0, PERF_COUNT(0));
 	metag_out32(0, PERF_COUNT(1));
 
-	for_each_possible_cpu(cpu) {
-		struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
+	cpuhp_setup_state(CPUHP_AP_PERF_METAG_STARTING,
+			  "AP_PERF_METAG_STARTING", metag_pmu_starting_cpu,
+			  NULL);
 
-		memset(cpuc, 0, sizeof(struct cpu_hw_events));
-		raw_spin_lock_init(&cpuc->pmu_lock);
-	}
-
-	register_cpu_notifier(&metag_pmu_notifier);
 	ret = perf_pmu_register(&pmu, metag_pmu->name, PERF_TYPE_RAW);
-out:
+	if (ret)
+		cpuhp_remove_state_nocalls(CPUHP_AP_PERF_METAG_STARTING);
 	return ret;
 }
 early_initcall(init_hw_perf_events);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 0a36e61..5c66102 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
+	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
-- 
2.8.1

[tip:smp/hotplug] metag/perf: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  04d045a681122c806e4f17754ed48e2b19af5d80
Gitweb:     http://git.kernel.org/tip/04d045a681122c806e4f17754ed48e2b19af5d80
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:49 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:27 +0200

metag/perf: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-metag@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.717395164@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/metag/kernel/perf/perf_event.c | 26 +++++++-------------------
 include/linux/cpuhotplug.h          |  1 +
 2 files changed, 8 insertions(+), 19 deletions(-)

diff --git a/arch/metag/kernel/perf/perf_event.c b/arch/metag/kernel/perf/perf_event.c
index 33a365f..052cba2 100644
--- a/arch/metag/kernel/perf/perf_event.c
+++ b/arch/metag/kernel/perf/perf_event.c
@@ -806,25 +806,16 @@ static struct metag_pmu _metag_pmu = {
 };
 
 /* PMU CPU hotplug notifier */
-static int metag_pmu_cpu_notify(struct notifier_block *b, unsigned long action,
-				void *hcpu)
+static int metag_pmu_starting_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned int)hcpu;
 	struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
 
-	if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING)
-		return NOTIFY_DONE;
-
 	memset(cpuc, 0, sizeof(struct cpu_hw_events));
 	raw_spin_lock_init(&cpuc->pmu_lock);
 
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block metag_pmu_notifier = {
-	.notifier_call = metag_pmu_cpu_notify,
-};
-
 /* PMU Initialisation */
 static int __init init_hw_perf_events(void)
 {
@@ -876,16 +867,13 @@ static int __init init_hw_perf_events(void)
 	metag_out32(0, PERF_COUNT(0));
 	metag_out32(0, PERF_COUNT(1));
 
-	for_each_possible_cpu(cpu) {
-		struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
+	cpuhp_setup_state(CPUHP_AP_PERF_METAG_STARTING,
+			  "AP_PERF_METAG_STARTING", metag_pmu_starting_cpu,
+			  NULL);
 
-		memset(cpuc, 0, sizeof(struct cpu_hw_events));
-		raw_spin_lock_init(&cpuc->pmu_lock);
-	}
-
-	register_cpu_notifier(&metag_pmu_notifier);
 	ret = perf_pmu_register(&pmu, metag_pmu->name, PERF_TYPE_RAW);
-out:
+	if (ret)
+		cpuhp_remove_state_nocalls(CPUHP_AP_PERF_METAG_STARTING);
 	return ret;
 }
 early_initcall(init_hw_perf_events);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 88986c9..990052c 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -34,6 +34,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CQM_STARTING,
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
+	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,

[patch V2 47/67] arm/l2c: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0046-arm-l2c-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2754 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Brad Mouring <brad.mouring@ni.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm/mm/cache-l2x0.c   | 27 +++++++++++++--------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index c61996c..cc12905 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -597,17 +597,16 @@ static void l2c310_configure(void __iomem *base)
 			      L310_POWER_CTRL);
 }
 
-static int l2c310_cpu_enable_flz(struct notifier_block *nb, unsigned long act, void *data)
+static int l2c310_starting_cpu(unsigned int cpu)
 {
-	switch (act & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
-		break;
-	case CPU_DYING:
-		set_auxcr(get_auxcr() & ~(BIT(3) | BIT(2) | BIT(1)));
-		break;
-	}
-	return NOTIFY_OK;
+	set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
+	return 0;
+}
+
+static int l2c310_dying_cpu(unsigned int cpu)
+{
+	set_auxcr(get_auxcr() & ~(BIT(3) | BIT(2) | BIT(1)));
+	return 0;
 }
 
 static void __init l2c310_enable(void __iomem *base, unsigned num_lock)
@@ -678,10 +677,10 @@ static void __init l2c310_enable(void __iomem *base, unsigned num_lock)
 			power_ctrl & L310_STNDBY_MODE_EN ? "en" : "dis");
 	}
 
-	if (aux & L310_AUX_CTRL_FULL_LINE_ZERO) {
-		set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
-		cpu_notifier(l2c310_cpu_enable_flz, 0);
-	}
+	if (aux & L310_AUX_CTRL_FULL_LINE_ZERO)
+		cpuhp_setup_state(CPUHP_AP_ARM_L2X0_STARTING,
+				  "AP_ARM_L2X0_STARTING", l2c310_starting_cpu,
+				  l2c310_dying_cpu);
 }
 
 static void __init l2c310_fixup(void __iomem *base, u32 cache_id,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 5c66102..1876377 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -38,6 +38,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
-- 
2.8.1

[tip:smp/hotplug] arm/l2c: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  9eeb2264775f0b60fcff3d186bf54c90b0191650
Gitweb:     http://git.kernel.org/tip/9eeb2264775f0b60fcff3d186bf54c90b0191650
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:50 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:28 +0200

arm/l2c: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Brad Mouring <brad.mouring@ni.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.801270887@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm/mm/cache-l2x0.c   | 27 +++++++++++++--------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index c61996c..cc12905 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -597,17 +597,16 @@ static void l2c310_configure(void __iomem *base)
 			      L310_POWER_CTRL);
 }
 
-static int l2c310_cpu_enable_flz(struct notifier_block *nb, unsigned long act, void *data)
+static int l2c310_starting_cpu(unsigned int cpu)
 {
-	switch (act & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
-		break;
-	case CPU_DYING:
-		set_auxcr(get_auxcr() & ~(BIT(3) | BIT(2) | BIT(1)));
-		break;
-	}
-	return NOTIFY_OK;
+	set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
+	return 0;
+}
+
+static int l2c310_dying_cpu(unsigned int cpu)
+{
+	set_auxcr(get_auxcr() & ~(BIT(3) | BIT(2) | BIT(1)));
+	return 0;
 }
 
 static void __init l2c310_enable(void __iomem *base, unsigned num_lock)
@@ -678,10 +677,10 @@ static void __init l2c310_enable(void __iomem *base, unsigned num_lock)
 			power_ctrl & L310_STNDBY_MODE_EN ? "en" : "dis");
 	}
 
-	if (aux & L310_AUX_CTRL_FULL_LINE_ZERO) {
-		set_auxcr(get_auxcr() | BIT(3) | BIT(2) | BIT(1));
-		cpu_notifier(l2c310_cpu_enable_flz, 0);
-	}
+	if (aux & L310_AUX_CTRL_FULL_LINE_ZERO)
+		cpuhp_setup_state(CPUHP_AP_ARM_L2X0_STARTING,
+				  "AP_ARM_L2X0_STARTING", l2c310_starting_cpu,
+				  l2c310_dying_cpu);
 }
 
 static void __init l2c310_fixup(void __iomem *base, u32 cache_id,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 990052c..2057e64 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -37,6 +37,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
+	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,

[patch V2 48/67] arm/twd: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0047-arm-twd-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2923 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine. The callbacks won't be invoked on
already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Heiko Stuebner <heiko@sntech.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm/kernel/smp_twd.c  | 31 +++++++++++--------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 12 insertions(+), 20 deletions(-)

diff --git a/arch/arm/kernel/smp_twd.c b/arch/arm/kernel/smp_twd.c
index b6ec65e..02d5e5e 100644
--- a/arch/arm/kernel/smp_twd.c
+++ b/arch/arm/kernel/smp_twd.c
@@ -310,24 +310,17 @@ static void twd_timer_setup(void)
 	enable_percpu_irq(clk->irq, 0);
 }
 
-static int twd_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+static int twd_timer_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		twd_timer_setup();
-		break;
-	case CPU_DYING:
-		twd_timer_stop();
-		break;
-	}
-
-	return NOTIFY_OK;
+	twd_timer_setup();
+	return 0;
 }
 
-static struct notifier_block twd_timer_cpu_nb = {
-	.notifier_call = twd_timer_cpu_notify,
-};
+static int twd_timer_dying_cpu(unsigned int cpu)
+{
+	twd_timer_stop();
+	return 0;
+}
 
 static int __init twd_local_timer_common_register(struct device_node *np)
 {
@@ -345,9 +338,9 @@ static int __init twd_local_timer_common_register(struct device_node *np)
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&twd_timer_cpu_nb);
-	if (err)
-		goto out_irq;
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_TWD_STARTING,
+				  "AP_ARM_TWD_STARTING",
+				  twd_timer_starting_cpu, twd_timer_dying_cpu);
 
 	twd_get_clock(np);
 	if (!of_property_read_bool(np, "always-on"))
@@ -365,8 +358,6 @@ static int __init twd_local_timer_common_register(struct device_node *np)
 
 	return 0;
 
-out_irq:
-	free_percpu_irq(twd_ppi, twd_evt);
 out_free:
 	iounmap(twd_base);
 	twd_base = NULL;
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 1876377..86ac982 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -41,6 +41,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
-- 
2.8.1

[tip:smp/hotplug] arm/twd: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  26b8768868c090151aa8c337773d429e0d5ca727
Gitweb:     http://git.kernel.org/tip/26b8768868c090151aa8c337773d429e0d5ca727
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:51 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:28 +0200

arm/twd: Convert to hotplug state machine

Install the callbacks via the state machine. The callbacks won't be invoked on
already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Heiko Stuebner <heiko@sntech.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153336.881124821@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm/kernel/smp_twd.c  | 31 +++++++++++--------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 12 insertions(+), 20 deletions(-)

diff --git a/arch/arm/kernel/smp_twd.c b/arch/arm/kernel/smp_twd.c
index b6ec65e..02d5e5e 100644
--- a/arch/arm/kernel/smp_twd.c
+++ b/arch/arm/kernel/smp_twd.c
@@ -310,24 +310,17 @@ static void twd_timer_setup(void)
 	enable_percpu_irq(clk->irq, 0);
 }
 
-static int twd_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+static int twd_timer_starting_cpu(unsigned int cpu)
 {
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		twd_timer_setup();
-		break;
-	case CPU_DYING:
-		twd_timer_stop();
-		break;
-	}
-
-	return NOTIFY_OK;
+	twd_timer_setup();
+	return 0;
 }
 
-static struct notifier_block twd_timer_cpu_nb = {
-	.notifier_call = twd_timer_cpu_notify,
-};
+static int twd_timer_dying_cpu(unsigned int cpu)
+{
+	twd_timer_stop();
+	return 0;
+}
 
 static int __init twd_local_timer_common_register(struct device_node *np)
 {
@@ -345,9 +338,9 @@ static int __init twd_local_timer_common_register(struct device_node *np)
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&twd_timer_cpu_nb);
-	if (err)
-		goto out_irq;
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM_TWD_STARTING,
+				  "AP_ARM_TWD_STARTING",
+				  twd_timer_starting_cpu, twd_timer_dying_cpu);
 
 	twd_get_clock(np);
 	if (!of_property_read_bool(np, "always-on"))
@@ -365,8 +358,6 @@ static int __init twd_local_timer_common_register(struct device_node *np)
 
 	return 0;
 
-out_irq:
-	free_percpu_irq(twd_ppi, twd_evt);
 out_free:
 	iounmap(twd_base);
 	twd_base = NULL;
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 2057e64..ad48881 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -40,6 +40,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,

[patch V2 49/67] arm/xen: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0048-arm-xen-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3465 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

The get_cpu() in xen_starting_cpu() boils down to preempt_disable() since
we already know the CPU we run on. Disabling preemption shouldn't be required
here from what I see since it we don't switch CPUs while invoking the function.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: xen-devel@lists.xenproject.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm/xen/enlighten.c   | 41 +++++++++++------------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 12 insertions(+), 30 deletions(-)

diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c
index 75cd734..d822e23 100644
--- a/arch/arm/xen/enlighten.c
+++ b/arch/arm/xen/enlighten.c
@@ -161,12 +161,11 @@ static struct notifier_block xen_pvclock_gtod_notifier = {
 	.notifier_call = xen_pvclock_gtod_notify,
 };
 
-static void xen_percpu_init(void)
+static int xen_starting_cpu(unsigned int cpu)
 {
 	struct vcpu_register_vcpu_info info;
 	struct vcpu_info *vcpup;
 	int err;
-	int cpu = get_cpu();
 
 	/* 
 	 * VCPUOP_register_vcpu_info cannot be called twice for the same
@@ -190,7 +189,13 @@ static void xen_percpu_init(void)
 
 after_register_vcpu_info:
 	enable_percpu_irq(xen_events_irq, 0);
-	put_cpu();
+	return 0;
+}
+
+static int xen_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(xen_events_irq);
+	return 0;
 }
 
 static void xen_restart(enum reboot_mode reboot_mode, const char *cmd)
@@ -209,28 +214,6 @@ static void xen_power_off(void)
 	BUG_ON(rc);
 }
 
-static int xen_cpu_notification(struct notifier_block *self,
-				unsigned long action,
-				void *hcpu)
-{
-	switch (action) {
-	case CPU_STARTING:
-		xen_percpu_init();
-		break;
-	case CPU_DYING:
-		disable_percpu_irq(xen_events_irq);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block xen_cpu_notifier = {
-	.notifier_call = xen_cpu_notification,
-};
-
 static irqreturn_t xen_arm_callback(int irq, void *arg)
 {
 	xen_hvm_evtchn_do_upcall();
@@ -351,16 +334,14 @@ static int __init xen_guest_init(void)
 		return -EINVAL;
 	}
 
-	xen_percpu_init();
-
-	register_cpu_notifier(&xen_cpu_notifier);
-
 	pv_time_ops.steal_clock = xen_stolen_accounting;
 	static_key_slow_inc(&paravirt_steal_enabled);
 	if (xen_initial_domain())
 		pvclock_gtod_register_notifier(&xen_pvclock_gtod_notifier);
 
-	return 0;
+	return cpuhp_setup_state(CPUHP_AP_ARM_XEN_STARTING,
+				 "AP_ARM_XEN_STARTING", xen_starting_cpu,
+				 xen_dying_cpu);
 }
 early_initcall(xen_guest_init);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 86ac982..f986963 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -48,6 +48,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
+	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

Re: [patch V2 49/67] arm/xen: Convert to hotplug state machine

[Stefano Stabellini]

On Wed, 13 Jul 2016, Anna-Maria Gleixner wrote:
> From: Richard Cochran <rcochran@linutronix.de>
> 
> Install the callbacks via the state machine and let the core invoke
> the callbacks on the already online CPUs.
> 
> The get_cpu() in xen_starting_cpu() boils down to preempt_disable() since
> we already know the CPU we run on. Disabling preemption shouldn't be required
> here from what I see since it we don't switch CPUs while invoking the function.
> 
> Signed-off-by: Richard Cochran <rcochran@linutronix.de>
> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Russell King <linux@armlinux.org.uk>
> Cc: Stefano Stabellini <sstabellini@kernel.org>
> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: linux-arm-kernel@lists.infradead.org
> Cc: xen-devel@lists.xenproject.org
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>

Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>


>  arch/arm/xen/enlighten.c   | 41 +++++++++++------------------------------
>  include/linux/cpuhotplug.h |  1 +
>  2 files changed, 12 insertions(+), 30 deletions(-)
> 
> diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c
> index 75cd734..d822e23 100644
> --- a/arch/arm/xen/enlighten.c
> +++ b/arch/arm/xen/enlighten.c
> @@ -161,12 +161,11 @@ static struct notifier_block xen_pvclock_gtod_notifier = {
>  	.notifier_call = xen_pvclock_gtod_notify,
>  };
>  
> -static void xen_percpu_init(void)
> +static int xen_starting_cpu(unsigned int cpu)
>  {
>  	struct vcpu_register_vcpu_info info;
>  	struct vcpu_info *vcpup;
>  	int err;
> -	int cpu = get_cpu();
>  
>  	/* 
>  	 * VCPUOP_register_vcpu_info cannot be called twice for the same
> @@ -190,7 +189,13 @@ static void xen_percpu_init(void)
>  
>  after_register_vcpu_info:
>  	enable_percpu_irq(xen_events_irq, 0);
> -	put_cpu();
> +	return 0;
> +}
> +
> +static int xen_dying_cpu(unsigned int cpu)
> +{
> +	disable_percpu_irq(xen_events_irq);
> +	return 0;
>  }
>  
>  static void xen_restart(enum reboot_mode reboot_mode, const char *cmd)
> @@ -209,28 +214,6 @@ static void xen_power_off(void)
>  	BUG_ON(rc);
>  }
>  
> -static int xen_cpu_notification(struct notifier_block *self,
> -				unsigned long action,
> -				void *hcpu)
> -{
> -	switch (action) {
> -	case CPU_STARTING:
> -		xen_percpu_init();
> -		break;
> -	case CPU_DYING:
> -		disable_percpu_irq(xen_events_irq);
> -		break;
> -	default:
> -		break;
> -	}
> -
> -	return NOTIFY_OK;
> -}
> -
> -static struct notifier_block xen_cpu_notifier = {
> -	.notifier_call = xen_cpu_notification,
> -};
> -
>  static irqreturn_t xen_arm_callback(int irq, void *arg)
>  {
>  	xen_hvm_evtchn_do_upcall();
> @@ -351,16 +334,14 @@ static int __init xen_guest_init(void)
>  		return -EINVAL;
>  	}
>  
> -	xen_percpu_init();
> -
> -	register_cpu_notifier(&xen_cpu_notifier);
> -
>  	pv_time_ops.steal_clock = xen_stolen_accounting;
>  	static_key_slow_inc(&paravirt_steal_enabled);
>  	if (xen_initial_domain())
>  		pvclock_gtod_register_notifier(&xen_pvclock_gtod_notifier);
>  
> -	return 0;
> +	return cpuhp_setup_state(CPUHP_AP_ARM_XEN_STARTING,
> +				 "AP_ARM_XEN_STARTING", xen_starting_cpu,
> +				 xen_dying_cpu);
>  }
>  early_initcall(xen_guest_init);
>  
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index 86ac982..f986963 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -48,6 +48,7 @@ enum cpuhp_state {
>  	CPUHP_AP_KVM_STARTING,
>  	CPUHP_AP_KVM_ARM_VGIC_STARTING,
>  	CPUHP_AP_KVM_ARM_TIMER_STARTING,
> +	CPUHP_AP_ARM_XEN_STARTING,
>  	CPUHP_AP_LEDTRIG_STARTING,
>  	CPUHP_AP_NOTIFY_STARTING,
>  	CPUHP_AP_ONLINE,
> -- 
> 2.8.1
> 
> 
> 

[tip:smp/hotplug] arm/xen: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  4761adb6f49048cddd65553a71ea2354b2a838be
Gitweb:     http://git.kernel.org/tip/4761adb6f49048cddd65553a71ea2354b2a838be
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:52 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:28 +0200

arm/xen: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

The get_cpu() in xen_starting_cpu() boils down to preempt_disable() since
we already know the CPU we run on. Disabling preemption shouldn't be required
here from what I see since it we don't switch CPUs while invoking the function.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Cc: xen-devel@lists.xenproject.org
Link: http://lkml.kernel.org/r/20160713153336.971559670@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm/xen/enlighten.c   | 41 +++++++++++------------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 12 insertions(+), 30 deletions(-)

diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c
index 75cd734..d822e23 100644
--- a/arch/arm/xen/enlighten.c
+++ b/arch/arm/xen/enlighten.c
@@ -161,12 +161,11 @@ static struct notifier_block xen_pvclock_gtod_notifier = {
 	.notifier_call = xen_pvclock_gtod_notify,
 };
 
-static void xen_percpu_init(void)
+static int xen_starting_cpu(unsigned int cpu)
 {
 	struct vcpu_register_vcpu_info info;
 	struct vcpu_info *vcpup;
 	int err;
-	int cpu = get_cpu();
 
 	/* 
 	 * VCPUOP_register_vcpu_info cannot be called twice for the same
@@ -190,7 +189,13 @@ static void xen_percpu_init(void)
 
 after_register_vcpu_info:
 	enable_percpu_irq(xen_events_irq, 0);
-	put_cpu();
+	return 0;
+}
+
+static int xen_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(xen_events_irq);
+	return 0;
 }
 
 static void xen_restart(enum reboot_mode reboot_mode, const char *cmd)
@@ -209,28 +214,6 @@ static void xen_power_off(void)
 	BUG_ON(rc);
 }
 
-static int xen_cpu_notification(struct notifier_block *self,
-				unsigned long action,
-				void *hcpu)
-{
-	switch (action) {
-	case CPU_STARTING:
-		xen_percpu_init();
-		break;
-	case CPU_DYING:
-		disable_percpu_irq(xen_events_irq);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block xen_cpu_notifier = {
-	.notifier_call = xen_cpu_notification,
-};
-
 static irqreturn_t xen_arm_callback(int irq, void *arg)
 {
 	xen_hvm_evtchn_do_upcall();
@@ -351,16 +334,14 @@ static int __init xen_guest_init(void)
 		return -EINVAL;
 	}
 
-	xen_percpu_init();
-
-	register_cpu_notifier(&xen_cpu_notifier);
-
 	pv_time_ops.steal_clock = xen_stolen_accounting;
 	static_key_slow_inc(&paravirt_steal_enabled);
 	if (xen_initial_domain())
 		pvclock_gtod_register_notifier(&xen_pvclock_gtod_notifier);
 
-	return 0;
+	return cpuhp_setup_state(CPUHP_AP_ARM_XEN_STARTING,
+				 "AP_ARM_XEN_STARTING", xen_starting_cpu,
+				 xen_dying_cpu);
 }
 early_initcall(xen_guest_init);
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ad48881..4c02b52 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -47,6 +47,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
+	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 50/67] MIPS/Loongson-3: Convert oprofile to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0049-MIPS-Loongson-3-Convert-oprofile-to-hotplug-state-ma.patch --]
[-- Type: text/plain, Size: 2985 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Robert Richter <rric@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-mips@linux-mips.org
Cc: oprofile-list@lists.sf.net
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/mips/oprofile/op_model_loongson3.c | 35 +++++++++++++--------------------
 include/linux/cpuhotplug.h              |  1 +
 2 files changed, 15 insertions(+), 21 deletions(-)

diff --git a/arch/mips/oprofile/op_model_loongson3.c b/arch/mips/oprofile/op_model_loongson3.c
index 8bcf7fc..85f3ee4 100644
--- a/arch/mips/oprofile/op_model_loongson3.c
+++ b/arch/mips/oprofile/op_model_loongson3.c
@@ -168,33 +168,26 @@ static int loongson3_perfcount_handler(void)
 	return handled;
 }
 
-static int loongson3_cpu_callback(struct notifier_block *nfb,
-	unsigned long action, void *hcpu)
+static int loongson3_starting_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		write_c0_perflo1(reg.control1);
-		write_c0_perflo2(reg.control2);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		write_c0_perflo1(0xc0000000);
-		write_c0_perflo2(0x40000000);
-		break;
-	}
-
-	return NOTIFY_OK;
+	write_c0_perflo1(reg.control1);
+	write_c0_perflo2(reg.control2);
+	return 0;
 }
 
-static struct notifier_block loongson3_notifier_block = {
-	.notifier_call = loongson3_cpu_callback
-};
+static int loongson3_dying_cpu(unsigned int cpu)
+{
+	write_c0_perflo1(0xc0000000);
+	write_c0_perflo2(0x40000000);
+	return 0;
+}
 
 static int __init loongson3_init(void)
 {
 	on_each_cpu(reset_counters, NULL, 1);
-	register_hotcpu_notifier(&loongson3_notifier_block);
+	cpuhp_setup_state_nocalls(CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
+				  "AP_MIPS_OP_LOONGSON3_STARTING",
+				  loongson3_starting_cpu, loongson3_dying_cpu);
 	save_perf_irq = perf_irq;
 	perf_irq = loongson3_perfcount_handler;
 
@@ -204,7 +197,7 @@ static int __init loongson3_init(void)
 static void loongson3_exit(void)
 {
 	on_each_cpu(reset_counters, NULL, 1);
-	unregister_hotcpu_notifier(&loongson3_notifier_block);
+	cpuhp_remove_state_nocalls(CPUHP_AP_MIPS_OP_LOONGSON3_STARTING);
 	perf_irq = save_perf_irq;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f986963..18dbf7b 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -36,6 +36,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_PERF_METAG_STARTING,
+	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_L2X0_STARTING,
-- 
2.8.1

[tip:smp/hotplug] MIPS/Loongson-3: Convert oprofile to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  65264e3bc3d479a87a3a9dbc6f3daee6439ce5c0
Gitweb:     http://git.kernel.org/tip/65264e3bc3d479a87a3a9dbc6f3daee6439ce5c0
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:53 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:29 +0200

MIPS/Loongson-3: Convert oprofile to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Robert Richter <rric@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-mips@linux-mips.org
Cc: oprofile-list@lists.sf.net
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.054827168@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/mips/oprofile/op_model_loongson3.c | 35 +++++++++++++--------------------
 include/linux/cpuhotplug.h              |  1 +
 2 files changed, 15 insertions(+), 21 deletions(-)

diff --git a/arch/mips/oprofile/op_model_loongson3.c b/arch/mips/oprofile/op_model_loongson3.c
index 8bcf7fc..85f3ee4 100644
--- a/arch/mips/oprofile/op_model_loongson3.c
+++ b/arch/mips/oprofile/op_model_loongson3.c
@@ -168,33 +168,26 @@ static int loongson3_perfcount_handler(void)
 	return handled;
 }
 
-static int loongson3_cpu_callback(struct notifier_block *nfb,
-	unsigned long action, void *hcpu)
+static int loongson3_starting_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		write_c0_perflo1(reg.control1);
-		write_c0_perflo2(reg.control2);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		write_c0_perflo1(0xc0000000);
-		write_c0_perflo2(0x40000000);
-		break;
-	}
-
-	return NOTIFY_OK;
+	write_c0_perflo1(reg.control1);
+	write_c0_perflo2(reg.control2);
+	return 0;
 }
 
-static struct notifier_block loongson3_notifier_block = {
-	.notifier_call = loongson3_cpu_callback
-};
+static int loongson3_dying_cpu(unsigned int cpu)
+{
+	write_c0_perflo1(0xc0000000);
+	write_c0_perflo2(0x40000000);
+	return 0;
+}
 
 static int __init loongson3_init(void)
 {
 	on_each_cpu(reset_counters, NULL, 1);
-	register_hotcpu_notifier(&loongson3_notifier_block);
+	cpuhp_setup_state_nocalls(CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
+				  "AP_MIPS_OP_LOONGSON3_STARTING",
+				  loongson3_starting_cpu, loongson3_dying_cpu);
 	save_perf_irq = perf_irq;
 	perf_irq = loongson3_perfcount_handler;
 
@@ -204,7 +197,7 @@ static int __init loongson3_init(void)
 static void loongson3_exit(void)
 {
 	on_each_cpu(reset_counters, NULL, 1);
-	unregister_hotcpu_notifier(&loongson3_notifier_block);
+	cpuhp_remove_state_nocalls(CPUHP_AP_MIPS_OP_LOONGSON3_STARTING);
 	perf_irq = save_perf_irq;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 4c02b52..ad26e35 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -35,6 +35,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_CSTATE_STARTING,
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_PERF_METAG_STARTING,
+	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_L2X0_STARTING,

[patch V2 51/67] hwtracing/coresight-etm3x: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0050-hwtracing-coresight-etm3x-Convert-to-hotplug-state-m.patch --]
[-- Type: text/plain, Size: 5150 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

This driver has an asymmetry of ONLINE code without any corresponding tear
down code. Otherwise, this is a straightforward conversion.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/hwtracing/coresight/coresight-etm3x.c |   90 ++++++++++++++------------
 include/linux/cpuhotplug.h                    |    1 
 2 files changed, 50 insertions(+), 41 deletions(-)

--- a/drivers/hwtracing/coresight/coresight-etm3x.c
+++ b/drivers/hwtracing/coresight/coresight-etm3x.c
@@ -51,6 +51,8 @@ module_param_named(boot_enable, boot_ena
 static int etm_count;
 static struct etm_drvdata *etmdrvdata[NR_CPUS];
 
+static enum cpuhp_state hp_online;
+
 /*
  * Memory mapped writes to clear os lock are not supported on some processors
  * and OS lock must be unlocked before any memory mapped access on such
@@ -481,8 +483,7 @@ static int etm_enable_sysfs(struct cores
 
 	/*
 	 * Configure the ETM only if the CPU is online.  If it isn't online
-	 * hw configuration will take place when 'CPU_STARTING' is received
-	 * in @etm_cpu_callback.
+	 * hw configuration will take place on the local CPU during bring up.
 	 */
 	if (cpu_online(drvdata->cpu)) {
 		ret = smp_call_function_single(drvdata->cpu,
@@ -641,47 +642,44 @@ static const struct coresight_ops etm_cs
 	.source_ops	= &etm_source_ops,
 };
 
-static int etm_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			    void *hcpu)
+static int etm_online_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	if (!etmdrvdata[cpu])
-		goto out;
+		return 0;
 
-	switch (action & (~CPU_TASKS_FROZEN)) {
-	case CPU_STARTING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (!etmdrvdata[cpu]->os_unlock) {
-			etm_os_unlock(etmdrvdata[cpu]);
-			etmdrvdata[cpu]->os_unlock = true;
-		}
-
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm_enable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable)
+		coresight_enable(etmdrvdata[cpu]->csdev);
+	return 0;
+}
 
-	case CPU_ONLINE:
-		if (etmdrvdata[cpu]->boot_enable &&
-		    !etmdrvdata[cpu]->sticky_enable)
-			coresight_enable(etmdrvdata[cpu]->csdev);
-		break;
+static int etm_starting_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
 
-	case CPU_DYING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm_disable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (!etmdrvdata[cpu]->os_unlock) {
+		etm_os_unlock(etmdrvdata[cpu]);
+		etmdrvdata[cpu]->os_unlock = true;
 	}
-out:
-	return NOTIFY_OK;
+
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm_enable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
 }
 
-static struct notifier_block etm_cpu_notifier = {
-	.notifier_call = etm_cpu_callback,
-};
+static int etm_dying_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
+
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm_disable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
+}
 
 static bool etm_arch_supported(u8 arch)
 {
@@ -806,9 +804,17 @@ static int etm_probe(struct amba_device
 				     etm_init_arch_data,  drvdata, 1))
 		dev_err(dev, "ETM arch init failed\n");
 
-	if (!etm_count++)
-		register_hotcpu_notifier(&etm_cpu_notifier);
-
+	if (!etm_count++) {
+		cpuhp_setup_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING,
+					  "AP_ARM_CORESIGHT_STARTING",
+					  etm_starting_cpu, etm_dying_cpu);
+		ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+						"AP_ARM_CORESIGHT_ONLINE",
+						etm_online_cpu, NULL);
+		if (ret < 0)
+			goto err_arch_supported;
+		hp_online = ret;
+	}
 	put_online_cpus();
 
 	if (etm_arch_supported(drvdata->arch) == false) {
@@ -839,7 +845,6 @@ static int etm_probe(struct amba_device
 
 	pm_runtime_put(&adev->dev);
 	dev_info(dev, "%s initialized\n", (char *)id->data);
-
 	if (boot_enable) {
 		coresight_enable(drvdata->csdev);
 		drvdata->boot_enable = true;
@@ -848,8 +853,11 @@ static int etm_probe(struct amba_device
 	return 0;
 
 err_arch_supported:
-	if (--etm_count == 0)
-		unregister_hotcpu_notifier(&etm_cpu_notifier);
+	if (--etm_count == 0) {
+		cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
+		if (hp_online)
+			cpuhp_remove_state_nocalls(hp_online);
+	}
 	return ret;
 }
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -50,6 +50,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
+	CPUHP_AP_ARM_CORESIGHT_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[tip:smp/hotplug] hwtracing/coresight-etm3x: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  2b5283d12f678391616718229aeabeca47497db5
Gitweb:     http://git.kernel.org/tip/2b5283d12f678391616718229aeabeca47497db5
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:54 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:29 +0200

hwtracing/coresight-etm3x: Convert to hotplug state machine

This driver has an asymmetry of ONLINE code without any corresponding tear
down code. Otherwise, this is a straightforward conversion.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.147128995@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/hwtracing/coresight/coresight-etm3x.c | 90 +++++++++++++++------------
 include/linux/cpuhotplug.h                    |  1 +
 2 files changed, 50 insertions(+), 41 deletions(-)

diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x.c
index d83ab82..2de4cad 100644
--- a/drivers/hwtracing/coresight/coresight-etm3x.c
+++ b/drivers/hwtracing/coresight/coresight-etm3x.c
@@ -51,6 +51,8 @@ module_param_named(boot_enable, boot_enable, int, S_IRUGO);
 static int etm_count;
 static struct etm_drvdata *etmdrvdata[NR_CPUS];
 
+static enum cpuhp_state hp_online;
+
 /*
  * Memory mapped writes to clear os lock are not supported on some processors
  * and OS lock must be unlocked before any memory mapped access on such
@@ -481,8 +483,7 @@ static int etm_enable_sysfs(struct coresight_device *csdev)
 
 	/*
 	 * Configure the ETM only if the CPU is online.  If it isn't online
-	 * hw configuration will take place when 'CPU_STARTING' is received
-	 * in @etm_cpu_callback.
+	 * hw configuration will take place on the local CPU during bring up.
 	 */
 	if (cpu_online(drvdata->cpu)) {
 		ret = smp_call_function_single(drvdata->cpu,
@@ -641,47 +642,44 @@ static const struct coresight_ops etm_cs_ops = {
 	.source_ops	= &etm_source_ops,
 };
 
-static int etm_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			    void *hcpu)
+static int etm_online_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	if (!etmdrvdata[cpu])
-		goto out;
+		return 0;
 
-	switch (action & (~CPU_TASKS_FROZEN)) {
-	case CPU_STARTING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (!etmdrvdata[cpu]->os_unlock) {
-			etm_os_unlock(etmdrvdata[cpu]);
-			etmdrvdata[cpu]->os_unlock = true;
-		}
-
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm_enable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable)
+		coresight_enable(etmdrvdata[cpu]->csdev);
+	return 0;
+}
 
-	case CPU_ONLINE:
-		if (etmdrvdata[cpu]->boot_enable &&
-		    !etmdrvdata[cpu]->sticky_enable)
-			coresight_enable(etmdrvdata[cpu]->csdev);
-		break;
+static int etm_starting_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
 
-	case CPU_DYING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm_disable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (!etmdrvdata[cpu]->os_unlock) {
+		etm_os_unlock(etmdrvdata[cpu]);
+		etmdrvdata[cpu]->os_unlock = true;
 	}
-out:
-	return NOTIFY_OK;
+
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm_enable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
 }
 
-static struct notifier_block etm_cpu_notifier = {
-	.notifier_call = etm_cpu_callback,
-};
+static int etm_dying_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
+
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm_disable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
+}
 
 static bool etm_arch_supported(u8 arch)
 {
@@ -806,9 +804,17 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id)
 				     etm_init_arch_data,  drvdata, 1))
 		dev_err(dev, "ETM arch init failed\n");
 
-	if (!etm_count++)
-		register_hotcpu_notifier(&etm_cpu_notifier);
-
+	if (!etm_count++) {
+		cpuhp_setup_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING,
+					  "AP_ARM_CORESIGHT_STARTING",
+					  etm_starting_cpu, etm_dying_cpu);
+		ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+						"AP_ARM_CORESIGHT_ONLINE",
+						etm_online_cpu, NULL);
+		if (ret < 0)
+			goto err_arch_supported;
+		hp_online = ret;
+	}
 	put_online_cpus();
 
 	if (etm_arch_supported(drvdata->arch) == false) {
@@ -839,7 +845,6 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id)
 
 	pm_runtime_put(&adev->dev);
 	dev_info(dev, "%s initialized\n", (char *)id->data);
-
 	if (boot_enable) {
 		coresight_enable(drvdata->csdev);
 		drvdata->boot_enable = true;
@@ -848,8 +853,11 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id)
 	return 0;
 
 err_arch_supported:
-	if (--etm_count == 0)
-		unregister_hotcpu_notifier(&etm_cpu_notifier);
+	if (--etm_count == 0) {
+		cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
+		if (hp_online)
+			cpuhp_remove_state_nocalls(hp_online);
+	}
 	return ret;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ad26e35..576ad23 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -49,6 +49,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
+	CPUHP_AP_ARM_CORESIGHT_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 52/67] hwtracing/coresight-etm4x: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0051-hwtracing-coresight-etm4x-Convert-to-hotplug-state-m.patch --]
[-- Type: text/plain, Size: 4480 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

This driver has an asymmetry of ONLINE code without any corresponding tear
down code. Otherwise, this is a straightforward conversion.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/hwtracing/coresight/coresight-etm4x.c |   85 ++++++++++++++------------
 include/linux/cpuhotplug.h                    |    1 
 2 files changed, 49 insertions(+), 37 deletions(-)

--- a/drivers/hwtracing/coresight/coresight-etm4x.c
+++ b/drivers/hwtracing/coresight/coresight-etm4x.c
@@ -48,6 +48,8 @@ static int etm4_count;
 static struct etmv4_drvdata *etmdrvdata[NR_CPUS];
 static void etm4_set_default(struct etmv4_config *config);
 
+static enum cpuhp_state hp_online;
+
 static void etm4_os_unlock(struct etmv4_drvdata *drvdata)
 {
 	/* Writing any value to ETMOSLAR unlocks the trace registers */
@@ -673,47 +675,44 @@ void etm4_config_trace_mode(struct etmv4
 	config->addr_acc[ETM_DEFAULT_ADDR_COMP + 1] = addr_acc;
 }
 
-static int etm4_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			    void *hcpu)
+static int etm4_online_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	if (!etmdrvdata[cpu])
-		goto out;
+		return 0;
 
-	switch (action & (~CPU_TASKS_FROZEN)) {
-	case CPU_STARTING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (!etmdrvdata[cpu]->os_unlock) {
-			etm4_os_unlock(etmdrvdata[cpu]);
-			etmdrvdata[cpu]->os_unlock = true;
-		}
-
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm4_enable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable)
+		coresight_enable(etmdrvdata[cpu]->csdev);
+	return 0;
+}
 
-	case CPU_ONLINE:
-		if (etmdrvdata[cpu]->boot_enable &&
-			!etmdrvdata[cpu]->sticky_enable)
-			coresight_enable(etmdrvdata[cpu]->csdev);
-		break;
+static int etm4_starting_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
 
-	case CPU_DYING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm4_disable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (!etmdrvdata[cpu]->os_unlock) {
+		etm4_os_unlock(etmdrvdata[cpu]);
+		etmdrvdata[cpu]->os_unlock = true;
 	}
-out:
-	return NOTIFY_OK;
+
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm4_enable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
 }
 
-static struct notifier_block etm4_cpu_notifier = {
-	.notifier_call = etm4_cpu_callback,
-};
+static int etm4_dying_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
+
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm4_disable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
+}
 
 static void etm4_init_trace_id(struct etmv4_drvdata *drvdata)
 {
@@ -767,8 +766,17 @@ static int etm4_probe(struct amba_device
 				etm4_init_arch_data,  drvdata, 1))
 		dev_err(dev, "ETM arch init failed\n");
 
-	if (!etm4_count++)
-		register_hotcpu_notifier(&etm4_cpu_notifier);
+	if (!etm4_count++) {
+		cpuhp_setup_state_nocalls(CPUHP_AP_ARM_CORESIGHT4_STARTING,
+					  "AP_ARM_CORESIGHT4_STARTING",
+					  etm4_starting_cpu, etm4_dying_cpu);
+		ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+						"AP_ARM_CORESIGHT4_ONLINE",
+						etm4_online_cpu, NULL);
+		if (ret < 0)
+			goto err_arch_supported;
+		hp_online = ret;
+	}
 
 	put_online_cpus();
 
@@ -809,8 +817,11 @@ static int etm4_probe(struct amba_device
 	return 0;
 
 err_arch_supported:
-	if (--etm4_count == 0)
-		unregister_hotcpu_notifier(&etm4_cpu_notifier);
+	if (--etm4_count == 0) {
+		cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT4_STARTING);
+		if (hp_online)
+			cpuhp_remove_state_nocalls(hp_online);
+	}
 	return ret;
 }
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -51,6 +51,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_ARM_CORESIGHT_STARTING,
+	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[tip:smp/hotplug] hwtracing/coresight-etm4x: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  58eb457be0283d9bfc3024eeefea33715391443f
Gitweb:     http://git.kernel.org/tip/58eb457be0283d9bfc3024eeefea33715391443f
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:55 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:30 +0200

hwtracing/coresight-etm4x: Convert to hotplug state machine

This driver has an asymmetry of ONLINE code without any corresponding tear
down code. Otherwise, this is a straightforward conversion.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Acked-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.228918408@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/hwtracing/coresight/coresight-etm4x.c | 87 +++++++++++++++------------
 include/linux/cpuhotplug.h                    |  1 +
 2 files changed, 50 insertions(+), 38 deletions(-)

diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x.c
index 462f0dc..1a5e0d1 100644
--- a/drivers/hwtracing/coresight/coresight-etm4x.c
+++ b/drivers/hwtracing/coresight/coresight-etm4x.c
@@ -48,6 +48,8 @@ static int etm4_count;
 static struct etmv4_drvdata *etmdrvdata[NR_CPUS];
 static void etm4_set_default(struct etmv4_config *config);
 
+static enum cpuhp_state hp_online;
+
 static void etm4_os_unlock(struct etmv4_drvdata *drvdata)
 {
 	/* Writing any value to ETMOSLAR unlocks the trace registers */
@@ -673,47 +675,44 @@ void etm4_config_trace_mode(struct etmv4_config *config)
 	config->addr_acc[ETM_DEFAULT_ADDR_COMP + 1] = addr_acc;
 }
 
-static int etm4_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			    void *hcpu)
+static int etm4_online_cpu(unsigned int cpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
 	if (!etmdrvdata[cpu])
-		goto out;
-
-	switch (action & (~CPU_TASKS_FROZEN)) {
-	case CPU_STARTING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (!etmdrvdata[cpu]->os_unlock) {
-			etm4_os_unlock(etmdrvdata[cpu]);
-			etmdrvdata[cpu]->os_unlock = true;
-		}
-
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm4_enable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+		return 0;
 
-	case CPU_ONLINE:
-		if (etmdrvdata[cpu]->boot_enable &&
-			!etmdrvdata[cpu]->sticky_enable)
-			coresight_enable(etmdrvdata[cpu]->csdev);
-		break;
+	if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable)
+		coresight_enable(etmdrvdata[cpu]->csdev);
+	return 0;
+}
 
-	case CPU_DYING:
-		spin_lock(&etmdrvdata[cpu]->spinlock);
-		if (local_read(&etmdrvdata[cpu]->mode))
-			etm4_disable_hw(etmdrvdata[cpu]);
-		spin_unlock(&etmdrvdata[cpu]->spinlock);
-		break;
+static int etm4_starting_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
+
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (!etmdrvdata[cpu]->os_unlock) {
+		etm4_os_unlock(etmdrvdata[cpu]);
+		etmdrvdata[cpu]->os_unlock = true;
 	}
-out:
-	return NOTIFY_OK;
+
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm4_enable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
 }
 
-static struct notifier_block etm4_cpu_notifier = {
-	.notifier_call = etm4_cpu_callback,
-};
+static int etm4_dying_cpu(unsigned int cpu)
+{
+	if (!etmdrvdata[cpu])
+		return 0;
+
+	spin_lock(&etmdrvdata[cpu]->spinlock);
+	if (local_read(&etmdrvdata[cpu]->mode))
+		etm4_disable_hw(etmdrvdata[cpu]);
+	spin_unlock(&etmdrvdata[cpu]->spinlock);
+	return 0;
+}
 
 static void etm4_init_trace_id(struct etmv4_drvdata *drvdata)
 {
@@ -767,8 +766,17 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id)
 				etm4_init_arch_data,  drvdata, 1))
 		dev_err(dev, "ETM arch init failed\n");
 
-	if (!etm4_count++)
-		register_hotcpu_notifier(&etm4_cpu_notifier);
+	if (!etm4_count++) {
+		cpuhp_setup_state_nocalls(CPUHP_AP_ARM_CORESIGHT4_STARTING,
+					  "AP_ARM_CORESIGHT4_STARTING",
+					  etm4_starting_cpu, etm4_dying_cpu);
+		ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+						"AP_ARM_CORESIGHT4_ONLINE",
+						etm4_online_cpu, NULL);
+		if (ret < 0)
+			goto err_arch_supported;
+		hp_online = ret;
+	}
 
 	put_online_cpus();
 
@@ -809,8 +817,11 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id)
 	return 0;
 
 err_arch_supported:
-	if (--etm4_count == 0)
-		unregister_hotcpu_notifier(&etm4_cpu_notifier);
+	if (--etm4_count == 0) {
+		cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT4_STARTING);
+		if (hp_online)
+			cpuhp_remove_state_nocalls(hp_online);
+	}
 	return ret;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 576ad23..6cede62 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -50,6 +50,7 @@ enum cpuhp_state {
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_ARM_CORESIGHT_STARTING,
+	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 53/67] arm64/armv8 deprecated: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0052-arm64-armv8-deprecated-Convert-to-hotplug-state-mach.patch --]
[-- Type: text/plain, Size: 3192 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Jisheng Zhang <jszhang@marvell.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shengjiu Wang <shengjiu.wang@freescale.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/arm64/kernel/armv8_deprecated.c | 22 +++++-----------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 6 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/kernel/armv8_deprecated.c b/arch/arm64/kernel/armv8_deprecated.c
index c37202c..9668c23 100644
--- a/arch/arm64/kernel/armv8_deprecated.c
+++ b/arch/arm64/kernel/armv8_deprecated.c
@@ -121,7 +121,7 @@ static int run_all_cpu_set_hw_mode(struct insn_emulation *insn, bool enable)
  *  0 		- If all the hooks ran successfully.
  * -EINVAL	- At least one hook is not supported by the CPU.
  */
-static int run_all_insn_set_hw_mode(unsigned long cpu)
+static int run_all_insn_set_hw_mode(unsigned int cpu)
 {
 	int rc = 0;
 	unsigned long flags;
@@ -131,7 +131,7 @@ static int run_all_insn_set_hw_mode(unsigned long cpu)
 	list_for_each_entry(insn, &insn_emulation, node) {
 		bool enable = (insn->current_mode == INSN_HW);
 		if (insn->ops->set_hw_mode && insn->ops->set_hw_mode(enable)) {
-			pr_warn("CPU[%ld] cannot support the emulation of %s",
+			pr_warn("CPU[%u] cannot support the emulation of %s",
 				cpu, insn->ops->name);
 			rc = -EINVAL;
 		}
@@ -617,20 +617,6 @@ static struct insn_emulation_ops setend_ops = {
 	.set_hw_mode = setend_set_hw_mode,
 };
 
-static int insn_cpu_hotplug_notify(struct notifier_block *b,
-			      unsigned long action, void *hcpu)
-{
-	int rc = 0;
-	if ((action & ~CPU_TASKS_FROZEN) == CPU_STARTING)
-		rc = run_all_insn_set_hw_mode((unsigned long)hcpu);
-
-	return notifier_from_errno(rc);
-}
-
-static struct notifier_block insn_cpu_hotplug_notifier = {
-	.notifier_call = insn_cpu_hotplug_notify,
-};
-
 /*
  * Invoked as late_initcall, since not needed before init spawned.
  */
@@ -649,7 +635,9 @@ static int __init armv8_deprecated_init(void)
 			pr_info("setend instruction emulation is not supported on the system");
 	}
 
-	register_cpu_notifier(&insn_cpu_hotplug_notifier);
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM64_ISNDEP_STARTING,
+				  "AP_ARM64_ISNDEP_STARTING",
+				  run_all_insn_set_hw_mode, NULL);
 	register_insn_emulation_sysctl(ctl_abi);
 
 	return 0;
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 9f9098e..da94513 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -52,6 +52,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_ARM_CORESIGHT_STARTING,
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
+	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
-- 
2.8.1

[tip:smp/hotplug] arm64/armv8 deprecated: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  27c01a8c73e15b2135b96a1866e1dbeb797e4806
Gitweb:     http://git.kernel.org/tip/27c01a8c73e15b2135b96a1866e1dbeb797e4806
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:56 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:30 +0200

arm64/armv8 deprecated: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Jisheng Zhang <jszhang@marvell.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shengjiu Wang <shengjiu.wang@freescale.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.311115906@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arm64/kernel/armv8_deprecated.c | 22 +++++-----------------
 include/linux/cpuhotplug.h           |  1 +
 2 files changed, 6 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/kernel/armv8_deprecated.c b/arch/arm64/kernel/armv8_deprecated.c
index c37202c..9668c23 100644
--- a/arch/arm64/kernel/armv8_deprecated.c
+++ b/arch/arm64/kernel/armv8_deprecated.c
@@ -121,7 +121,7 @@ static int run_all_cpu_set_hw_mode(struct insn_emulation *insn, bool enable)
  *  0 		- If all the hooks ran successfully.
  * -EINVAL	- At least one hook is not supported by the CPU.
  */
-static int run_all_insn_set_hw_mode(unsigned long cpu)
+static int run_all_insn_set_hw_mode(unsigned int cpu)
 {
 	int rc = 0;
 	unsigned long flags;
@@ -131,7 +131,7 @@ static int run_all_insn_set_hw_mode(unsigned long cpu)
 	list_for_each_entry(insn, &insn_emulation, node) {
 		bool enable = (insn->current_mode == INSN_HW);
 		if (insn->ops->set_hw_mode && insn->ops->set_hw_mode(enable)) {
-			pr_warn("CPU[%ld] cannot support the emulation of %s",
+			pr_warn("CPU[%u] cannot support the emulation of %s",
 				cpu, insn->ops->name);
 			rc = -EINVAL;
 		}
@@ -617,20 +617,6 @@ static struct insn_emulation_ops setend_ops = {
 	.set_hw_mode = setend_set_hw_mode,
 };
 
-static int insn_cpu_hotplug_notify(struct notifier_block *b,
-			      unsigned long action, void *hcpu)
-{
-	int rc = 0;
-	if ((action & ~CPU_TASKS_FROZEN) == CPU_STARTING)
-		rc = run_all_insn_set_hw_mode((unsigned long)hcpu);
-
-	return notifier_from_errno(rc);
-}
-
-static struct notifier_block insn_cpu_hotplug_notifier = {
-	.notifier_call = insn_cpu_hotplug_notify,
-};
-
 /*
  * Invoked as late_initcall, since not needed before init spawned.
  */
@@ -649,7 +635,9 @@ static int __init armv8_deprecated_init(void)
 			pr_info("setend instruction emulation is not supported on the system");
 	}
 
-	register_cpu_notifier(&insn_cpu_hotplug_notifier);
+	cpuhp_setup_state_nocalls(CPUHP_AP_ARM64_ISNDEP_STARTING,
+				  "AP_ARM64_ISNDEP_STARTING",
+				  run_all_insn_set_hw_mode, NULL);
 	register_insn_emulation_sysctl(ctl_abi);
 
 	return 0;
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6cede62..c5b96d0 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -51,6 +51,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_XEN_STARTING,
 	CPUHP_AP_ARM_CORESIGHT_STARTING,
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
+	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,

[patch V2 54/67] x86/tboot: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0053-x86-tboot-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2631 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Gang Wei <gang.wei@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ning Sun <ning.sun@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Richard L Maliszewski <richard.l.maliszewski@intel.com>
Cc: Shane Wang <shane.wang@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: tboot-devel@lists.sourceforge.net
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/kernel/tboot.c    | 25 ++++++++-----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 9 insertions(+), 17 deletions(-)

diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c
index 9b0185f..654f6c6 100644
--- a/arch/x86/kernel/tboot.c
+++ b/arch/x86/kernel/tboot.c
@@ -323,25 +323,16 @@ static int tboot_wait_for_aps(int num_aps)
 	return !(atomic_read((atomic_t *)&tboot->num_in_wfs) == num_aps);
 }
 
-static int tboot_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			      void *hcpu)
+static int tboot_dying_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_DYING:
-		atomic_inc(&ap_wfs_count);
-		if (num_online_cpus() == 1)
-			if (tboot_wait_for_aps(atomic_read(&ap_wfs_count)))
-				return NOTIFY_BAD;
-		break;
+	atomic_inc(&ap_wfs_count);
+	if (num_online_cpus() == 1) {
+		if (tboot_wait_for_aps(atomic_read(&ap_wfs_count)))
+			return -EBUSY;
 	}
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block tboot_cpu_notifier =
-{
-	.notifier_call = tboot_cpu_callback,
-};
-
 #ifdef CONFIG_DEBUG_FS
 
 #define TBOOT_LOG_UUID	{ 0x26, 0x25, 0x19, 0xc0, 0x30, 0x6b, 0xb4, 0x4d, \
@@ -417,8 +408,8 @@ static __init int tboot_late_init(void)
 	tboot_create_trampoline();
 
 	atomic_set(&ap_wfs_count, 0);
-	register_hotcpu_notifier(&tboot_cpu_notifier);
-
+	cpuhp_setup_state(CPUHP_AP_X86_TBOOT_DYING, "AP_X86_TBOOT_DYING", NULL,
+			  tboot_dying_cpu);
 #ifdef CONFIG_DEBUG_FS
 	debugfs_create_file("tboot_log", S_IRUSR,
 			arch_debugfs_dir, NULL, &tboot_log_fops);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index da94513..c1597d6 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -54,6 +54,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
+	CPUHP_AP_X86_TBOOT_DYING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
-- 
2.8.1

[tip:smp/hotplug] x86/tboot: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  ae6a8a2ed721ecdbc15d32b2089af5ed2190adc7
Gitweb:     http://git.kernel.org/tip/ae6a8a2ed721ecdbc15d32b2089af5ed2190adc7
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:57 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:40:30 +0200

x86/tboot: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Gang Wei <gang.wei@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ning Sun <ning.sun@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Richard L Maliszewski <richard.l.maliszewski@intel.com>
Cc: Shane Wang <shane.wang@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Cc: tboot-devel@lists.sourceforge.net
Link: http://lkml.kernel.org/r/20160713153337.400227322@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/tboot.c    | 25 ++++++++-----------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 9 insertions(+), 17 deletions(-)

diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c
index 9b0185f..654f6c6 100644
--- a/arch/x86/kernel/tboot.c
+++ b/arch/x86/kernel/tboot.c
@@ -323,25 +323,16 @@ static int tboot_wait_for_aps(int num_aps)
 	return !(atomic_read((atomic_t *)&tboot->num_in_wfs) == num_aps);
 }
 
-static int tboot_cpu_callback(struct notifier_block *nfb, unsigned long action,
-			      void *hcpu)
+static int tboot_dying_cpu(unsigned int cpu)
 {
-	switch (action) {
-	case CPU_DYING:
-		atomic_inc(&ap_wfs_count);
-		if (num_online_cpus() == 1)
-			if (tboot_wait_for_aps(atomic_read(&ap_wfs_count)))
-				return NOTIFY_BAD;
-		break;
+	atomic_inc(&ap_wfs_count);
+	if (num_online_cpus() == 1) {
+		if (tboot_wait_for_aps(atomic_read(&ap_wfs_count)))
+			return -EBUSY;
 	}
-	return NOTIFY_OK;
+	return 0;
 }
 
-static struct notifier_block tboot_cpu_notifier =
-{
-	.notifier_call = tboot_cpu_callback,
-};
-
 #ifdef CONFIG_DEBUG_FS
 
 #define TBOOT_LOG_UUID	{ 0x26, 0x25, 0x19, 0xc0, 0x30, 0x6b, 0xb4, 0x4d, \
@@ -417,8 +408,8 @@ static __init int tboot_late_init(void)
 	tboot_create_trampoline();
 
 	atomic_set(&ap_wfs_count, 0);
-	register_hotcpu_notifier(&tboot_cpu_notifier);
-
+	cpuhp_setup_state(CPUHP_AP_X86_TBOOT_DYING, "AP_X86_TBOOT_DYING", NULL,
+			  tboot_dying_cpu);
 #ifdef CONFIG_DEBUG_FS
 	debugfs_create_file("tboot_log", S_IRUSR,
 			arch_debugfs_dir, NULL, &tboot_log_fops);
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index c5b96d0..070cc7f 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -53,6 +53,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
+	CPUHP_AP_X86_TBOOT_DYING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,

[patch V2 55/67] hrtimer: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0054-hrtimer-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3705 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Split out the clockevents callbacks instead of piggy packing them on
hrtimers.

This gets rid of a POST_DEAD user. See commit:

  54e88fad223c ("sched: Make sure timers have migrated before killing the migration_thread")

We just move the callback state to the proper place in the state machine.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |    1 +
 include/linux/hrtimer.h    |    7 +++++++
 kernel/cpu.c               |    5 +++++
 kernel/time/hrtimer.c      |   40 +++++-----------------------------------
 4 files changed, 18 insertions(+), 35 deletions(-)

--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -16,6 +16,7 @@ enum cpuhp_state {
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_POWER_NUMA_PREPARE,
+	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -494,4 +494,11 @@ extern void __init hrtimers_init(void);
 /* Show pending timers: */
 extern void sysrq_timer_list_show(void);
 
+int hrtimers_prepare_cpu(unsigned int cpu);
+#ifdef CONFIG_HOTPLUG_CPU
+int hrtimers_dead_cpu(unsigned int cpu);
+#else
+#define hrtimers_dead_cpu	NULL
+#endif
+
 #endif
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1190,6 +1190,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.startup = workqueue_prepare_cpu,
 		.teardown = NULL,
 	},
+	[CPUHP_HRTIMERS_PREPARE] = {
+		.name = "hrtimers prepare",
+		.startup = hrtimers_prepare_cpu,
+		.teardown = hrtimers_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1590,7 +1590,7 @@ SYSCALL_DEFINE2(nanosleep, struct timesp
 /*
  * Functions related to boot-time initialization:
  */
-static void init_hrtimers_cpu(int cpu)
+int hrtimers_prepare_cpu(unsigned int cpu)
 {
 	struct hrtimer_cpu_base *cpu_base = &per_cpu(hrtimer_bases, cpu);
 	int i;
@@ -1602,6 +1602,7 @@ static void init_hrtimers_cpu(int cpu)
 
 	cpu_base->cpu = cpu;
 	hrtimer_init_hres(cpu_base);
+	return 0;
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -1636,7 +1637,7 @@ static void migrate_hrtimer_list(struct
 	}
 }
 
-static void migrate_hrtimers(int scpu)
+int hrtimers_dead_cpu(unsigned int scpu)
 {
 	struct hrtimer_cpu_base *old_base, *new_base;
 	int i;
@@ -1665,45 +1666,14 @@ static void migrate_hrtimers(int scpu)
 	/* Check, if we got expired work to do */
 	__hrtimer_peek_ahead_timers();
 	local_irq_enable();
+	return 0;
 }
 
 #endif /* CONFIG_HOTPLUG_CPU */
 
-static int hrtimer_cpu_notify(struct notifier_block *self,
-					unsigned long action, void *hcpu)
-{
-	int scpu = (long)hcpu;
-
-	switch (action) {
-
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		init_hrtimers_cpu(scpu);
-		break;
-
-#ifdef CONFIG_HOTPLUG_CPU
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		migrate_hrtimers(scpu);
-		break;
-#endif
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block hrtimers_nb = {
-	.notifier_call = hrtimer_cpu_notify,
-};
-
 void __init hrtimers_init(void)
 {
-	hrtimer_cpu_notify(&hrtimers_nb, (unsigned long)CPU_UP_PREPARE,
-			  (void *)(long)smp_processor_id());
-	register_cpu_notifier(&hrtimers_nb);
+	hrtimers_prepare_cpu(smp_processor_id());
 }
 
 /**

[tip:smp/hotplug] hrtimer: Convert to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  27590dc17b34aedc4f3e14bd107ee59b9db9b0a6
Gitweb:     http://git.kernel.org/tip/27590dc17b34aedc4f3e14bd107ee59b9db9b0a6
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 15 Jul 2016 10:41:04 +0200
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:37 +0200

hrtimer: Convert to hotplug state machine

Split out the clockevents callbacks instead of piggybacking them on
hrtimers.

This gets rid of a POST_DEAD user. See commit:

  54e88fad223c ("sched: Make sure timers have migrated before killing the migration_thread")

We just move the callback state to the proper place in the state machine.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.485419196@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  1 +
 include/linux/hrtimer.h    |  7 +++++++
 kernel/cpu.c               |  5 +++++
 kernel/time/hrtimer.c      | 40 +++++-----------------------------------
 4 files changed, 18 insertions(+), 35 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 070cc7f..04ecc55 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -15,6 +15,7 @@ enum cpuhp_state {
 	CPUHP_X86_HPET_DEAD,
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
+	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index c98c653..5e00f80 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -494,4 +494,11 @@ extern void __init hrtimers_init(void);
 /* Show pending timers: */
 extern void sysrq_timer_list_show(void);
 
+int hrtimers_prepare_cpu(unsigned int cpu);
+#ifdef CONFIG_HOTPLUG_CPU
+int hrtimers_dead_cpu(unsigned int cpu);
+#else
+#define hrtimers_dead_cpu	NULL
+#endif
+
 #endif
diff --git a/kernel/cpu.c b/kernel/cpu.c
index af53f82..85500e7 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1190,6 +1190,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = workqueue_prepare_cpu,
 		.teardown = NULL,
 	},
+	[CPUHP_HRTIMERS_PREPARE] = {
+		.name = "hrtimers prepare",
+		.startup = hrtimers_prepare_cpu,
+		.teardown = hrtimers_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index d13c9ae..9ba7c82 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1590,7 +1590,7 @@ SYSCALL_DEFINE2(nanosleep, struct timespec __user *, rqtp,
 /*
  * Functions related to boot-time initialization:
  */
-static void init_hrtimers_cpu(int cpu)
+int hrtimers_prepare_cpu(unsigned int cpu)
 {
 	struct hrtimer_cpu_base *cpu_base = &per_cpu(hrtimer_bases, cpu);
 	int i;
@@ -1602,6 +1602,7 @@ static void init_hrtimers_cpu(int cpu)
 
 	cpu_base->cpu = cpu;
 	hrtimer_init_hres(cpu_base);
+	return 0;
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -1636,7 +1637,7 @@ static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
 	}
 }
 
-static void migrate_hrtimers(int scpu)
+int hrtimers_dead_cpu(unsigned int scpu)
 {
 	struct hrtimer_cpu_base *old_base, *new_base;
 	int i;
@@ -1665,45 +1666,14 @@ static void migrate_hrtimers(int scpu)
 	/* Check, if we got expired work to do */
 	__hrtimer_peek_ahead_timers();
 	local_irq_enable();
+	return 0;
 }
 
 #endif /* CONFIG_HOTPLUG_CPU */
 
-static int hrtimer_cpu_notify(struct notifier_block *self,
-					unsigned long action, void *hcpu)
-{
-	int scpu = (long)hcpu;
-
-	switch (action) {
-
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		init_hrtimers_cpu(scpu);
-		break;
-
-#ifdef CONFIG_HOTPLUG_CPU
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		migrate_hrtimers(scpu);
-		break;
-#endif
-
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block hrtimers_nb = {
-	.notifier_call = hrtimer_cpu_notify,
-};
-
 void __init hrtimers_init(void)
 {
-	hrtimer_cpu_notify(&hrtimers_nb, (unsigned long)CPU_UP_PREPARE,
-			  (void *)(long)smp_processor_id());
-	register_cpu_notifier(&hrtimers_nb);
+	hrtimers_prepare_cpu(smp_processor_id());
 }
 
 /**

[patch V2 57/67] profile: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0056-profile-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 6858 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs. A lot of code is removed because
the for-loop is used and create_hash_tables() is removed since its purpose
is covered by the startup / teardown hooks.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |   1 +
 kernel/profile.c           | 181 ++++++++++++++++-----------------------------
 2 files changed, 66 insertions(+), 116 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index e0f8bfd..a38fb6f 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -17,6 +17,7 @@ enum cpuhp_state {
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,
+	CPUHP_PROFILE_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
diff --git a/kernel/profile.c b/kernel/profile.c
index c2199e9..2dbccf2 100644
--- a/kernel/profile.c
+++ b/kernel/profile.c
@@ -328,68 +328,57 @@ out:
 	put_cpu();
 }
 
-static int profile_cpu_callback(struct notifier_block *info,
-					unsigned long action, void *__cpu)
+static int profile_dead_cpu(unsigned int cpu)
 {
-	int node, cpu = (unsigned long)__cpu;
 	struct page *page;
+	int i;
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		node = cpu_to_mem(cpu);
-		per_cpu(cpu_profile_flip, cpu) = 0;
-		if (!per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = __alloc_pages_node(node,
-					GFP_KERNEL | __GFP_ZERO,
-					0);
-			if (!page)
-				return notifier_from_errno(-ENOMEM);
-			per_cpu(cpu_profile_hits, cpu)[1] = page_address(page);
-		}
-		if (!per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = __alloc_pages_node(node,
-					GFP_KERNEL | __GFP_ZERO,
-					0);
-			if (!page)
-				goto out_free;
-			per_cpu(cpu_profile_hits, cpu)[0] = page_address(page);
-		}
-		break;
-out_free:
-		page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-		per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-		__free_page(page);
-		return notifier_from_errno(-ENOMEM);
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		if (prof_cpu_mask != NULL)
-			cpumask_set_cpu(cpu, prof_cpu_mask);
-		break;
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		if (prof_cpu_mask != NULL)
-			cpumask_clear_cpu(cpu, prof_cpu_mask);
-		if (per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[0]);
-			per_cpu(cpu_profile_hits, cpu)[0] = NULL;
+	if (prof_cpu_mask != NULL)
+		cpumask_clear_cpu(cpu, prof_cpu_mask);
+
+	for (i = 0; i < 2; i++) {
+		if (per_cpu(cpu_profile_hits, cpu)[i]) {
+			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[i]);
+			per_cpu(cpu_profile_hits, cpu)[i] = NULL;
 			__free_page(page);
 		}
-		if (per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-			per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-			__free_page(page);
+	}
+	return 0;
+}
+
+static int profile_prepare_cpu(unsigned int cpu)
+{
+	int i, node = cpu_to_mem(cpu);
+	struct page *page;
+
+	per_cpu(cpu_profile_flip, cpu) = 0;
+
+	for (i = 0; i < 2; i++) {
+		if (per_cpu(cpu_profile_hits, cpu)[i])
+			continue;
+
+		page = __alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0);
+		if (!page) {
+			profile_dead_cpu(cpu);
+			return -ENOMEM;
 		}
-		break;
+		per_cpu(cpu_profile_hits, cpu)[i] = page_address(page);
+
 	}
-	return NOTIFY_OK;
+	return 0;
+}
+
+static int profile_online_cpu(unsigned int cpu)
+{
+	if (prof_cpu_mask != NULL)
+		cpumask_set_cpu(cpu, prof_cpu_mask);
+
+	return 0;
 }
+
 #else /* !CONFIG_SMP */
 #define profile_flip_buffers()		do { } while (0)
 #define profile_discard_flip_buffers()	do { } while (0)
-#define profile_cpu_callback		NULL
 
 static void do_profile_hits(int type, void *__pc, unsigned int nr_hits)
 {
@@ -531,83 +520,43 @@ static const struct file_operations proc_profile_operations = {
 	.llseek		= default_llseek,
 };
 
-#ifdef CONFIG_SMP
-static void profile_nop(void *unused)
-{
-}
-
-static int create_hash_tables(void)
+int __ref create_proc_profile(void)
 {
-	int cpu;
-
-	for_each_online_cpu(cpu) {
-		int node = cpu_to_mem(cpu);
-		struct page *page;
-
-		page = __alloc_pages_node(node,
-				GFP_KERNEL | __GFP_ZERO | __GFP_THISNODE,
-				0);
-		if (!page)
-			goto out_cleanup;
-		per_cpu(cpu_profile_hits, cpu)[1]
-				= (struct profile_hit *)page_address(page);
-		page = __alloc_pages_node(node,
-				GFP_KERNEL | __GFP_ZERO | __GFP_THISNODE,
-				0);
-		if (!page)
-			goto out_cleanup;
-		per_cpu(cpu_profile_hits, cpu)[0]
-				= (struct profile_hit *)page_address(page);
-	}
-	return 0;
-out_cleanup:
-	prof_on = 0;
-	smp_mb();
-	on_each_cpu(profile_nop, NULL, 1);
-	for_each_online_cpu(cpu) {
-		struct page *page;
-
-		if (per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[0]);
-			per_cpu(cpu_profile_hits, cpu)[0] = NULL;
-			__free_page(page);
-		}
-		if (per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-			per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-			__free_page(page);
-		}
-	}
-	return -1;
-}
-#else
-#define create_hash_tables()			({ 0; })
+	struct proc_dir_entry *entry;
+#ifdef CONFIG_SMP
+	enum cpuhp_state online_state;
 #endif
 
-int __ref create_proc_profile(void) /* false positive from hotcpu_notifier */
-{
-	struct proc_dir_entry *entry;
 	int err = 0;
 
 	if (!prof_on)
 		return 0;
-
-	cpu_notifier_register_begin();
-
-	if (create_hash_tables()) {
-		err = -ENOMEM;
-		goto out;
-	}
-
+#ifdef CONFIG_SMP
+	err = cpuhp_setup_state(CPUHP_PROFILE_PREPARE, "PROFILE_PREPARE",
+				profile_prepare_cpu, profile_dead_cpu);
+	if (err)
+		return err;
+
+	err = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "AP_PROFILE_ONLINE",
+				profile_online_cpu, NULL);
+	if (err < 0)
+		goto err_state_prep;
+	online_state = err;
+	err = 0;
+#endif
 	entry = proc_create("profile", S_IWUSR | S_IRUGO,
 			    NULL, &proc_profile_operations);
 	if (!entry)
-		goto out;
+		goto err_state_onl;
 	proc_set_size(entry, (1 + prof_len) * sizeof(atomic_t));
-	__hotcpu_notifier(profile_cpu_callback, 0);
 
-out:
-	cpu_notifier_register_done();
+	return err;
+err_state_onl:
+#ifdef CONFIG_SMP
+	cpuhp_remove_state(online_state);
+err_state_prep:
+	cpuhp_remove_state(CPUHP_PROFILE_PREPARE);
+#endif
 	return err;
 }
 subsys_initcall(create_proc_profile);
-- 
2.8.1

[tip:smp/hotplug] profile: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  e722d8daafb974b9ad1bbaf42f384a5ea5929f5f
Gitweb:     http://git.kernel.org/tip/e722d8daafb974b9ad1bbaf42f384a5ea5929f5f
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:59 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:42 +0200

profile: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs. A lot of code is removed because
the for-loop is used and create_hash_tables() is removed since its purpose
is covered by the startup / teardown hooks.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.649867675@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |   1 +
 kernel/profile.c           | 181 ++++++++++++++++-----------------------------
 2 files changed, 66 insertions(+), 116 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 15d46d2..ace5ad0 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -16,6 +16,7 @@ enum cpuhp_state {
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_HRTIMERS_PREPARE,
+	CPUHP_PROFILE_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
diff --git a/kernel/profile.c b/kernel/profile.c
index c2199e9..2dbccf2 100644
--- a/kernel/profile.c
+++ b/kernel/profile.c
@@ -328,68 +328,57 @@ out:
 	put_cpu();
 }
 
-static int profile_cpu_callback(struct notifier_block *info,
-					unsigned long action, void *__cpu)
+static int profile_dead_cpu(unsigned int cpu)
 {
-	int node, cpu = (unsigned long)__cpu;
 	struct page *page;
+	int i;
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		node = cpu_to_mem(cpu);
-		per_cpu(cpu_profile_flip, cpu) = 0;
-		if (!per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = __alloc_pages_node(node,
-					GFP_KERNEL | __GFP_ZERO,
-					0);
-			if (!page)
-				return notifier_from_errno(-ENOMEM);
-			per_cpu(cpu_profile_hits, cpu)[1] = page_address(page);
-		}
-		if (!per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = __alloc_pages_node(node,
-					GFP_KERNEL | __GFP_ZERO,
-					0);
-			if (!page)
-				goto out_free;
-			per_cpu(cpu_profile_hits, cpu)[0] = page_address(page);
-		}
-		break;
-out_free:
-		page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-		per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-		__free_page(page);
-		return notifier_from_errno(-ENOMEM);
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		if (prof_cpu_mask != NULL)
-			cpumask_set_cpu(cpu, prof_cpu_mask);
-		break;
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		if (prof_cpu_mask != NULL)
-			cpumask_clear_cpu(cpu, prof_cpu_mask);
-		if (per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[0]);
-			per_cpu(cpu_profile_hits, cpu)[0] = NULL;
+	if (prof_cpu_mask != NULL)
+		cpumask_clear_cpu(cpu, prof_cpu_mask);
+
+	for (i = 0; i < 2; i++) {
+		if (per_cpu(cpu_profile_hits, cpu)[i]) {
+			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[i]);
+			per_cpu(cpu_profile_hits, cpu)[i] = NULL;
 			__free_page(page);
 		}
-		if (per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-			per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-			__free_page(page);
+	}
+	return 0;
+}
+
+static int profile_prepare_cpu(unsigned int cpu)
+{
+	int i, node = cpu_to_mem(cpu);
+	struct page *page;
+
+	per_cpu(cpu_profile_flip, cpu) = 0;
+
+	for (i = 0; i < 2; i++) {
+		if (per_cpu(cpu_profile_hits, cpu)[i])
+			continue;
+
+		page = __alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0);
+		if (!page) {
+			profile_dead_cpu(cpu);
+			return -ENOMEM;
 		}
-		break;
+		per_cpu(cpu_profile_hits, cpu)[i] = page_address(page);
+
 	}
-	return NOTIFY_OK;
+	return 0;
+}
+
+static int profile_online_cpu(unsigned int cpu)
+{
+	if (prof_cpu_mask != NULL)
+		cpumask_set_cpu(cpu, prof_cpu_mask);
+
+	return 0;
 }
+
 #else /* !CONFIG_SMP */
 #define profile_flip_buffers()		do { } while (0)
 #define profile_discard_flip_buffers()	do { } while (0)
-#define profile_cpu_callback		NULL
 
 static void do_profile_hits(int type, void *__pc, unsigned int nr_hits)
 {
@@ -531,83 +520,43 @@ static const struct file_operations proc_profile_operations = {
 	.llseek		= default_llseek,
 };
 
-#ifdef CONFIG_SMP
-static void profile_nop(void *unused)
-{
-}
-
-static int create_hash_tables(void)
+int __ref create_proc_profile(void)
 {
-	int cpu;
-
-	for_each_online_cpu(cpu) {
-		int node = cpu_to_mem(cpu);
-		struct page *page;
-
-		page = __alloc_pages_node(node,
-				GFP_KERNEL | __GFP_ZERO | __GFP_THISNODE,
-				0);
-		if (!page)
-			goto out_cleanup;
-		per_cpu(cpu_profile_hits, cpu)[1]
-				= (struct profile_hit *)page_address(page);
-		page = __alloc_pages_node(node,
-				GFP_KERNEL | __GFP_ZERO | __GFP_THISNODE,
-				0);
-		if (!page)
-			goto out_cleanup;
-		per_cpu(cpu_profile_hits, cpu)[0]
-				= (struct profile_hit *)page_address(page);
-	}
-	return 0;
-out_cleanup:
-	prof_on = 0;
-	smp_mb();
-	on_each_cpu(profile_nop, NULL, 1);
-	for_each_online_cpu(cpu) {
-		struct page *page;
-
-		if (per_cpu(cpu_profile_hits, cpu)[0]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[0]);
-			per_cpu(cpu_profile_hits, cpu)[0] = NULL;
-			__free_page(page);
-		}
-		if (per_cpu(cpu_profile_hits, cpu)[1]) {
-			page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
-			per_cpu(cpu_profile_hits, cpu)[1] = NULL;
-			__free_page(page);
-		}
-	}
-	return -1;
-}
-#else
-#define create_hash_tables()			({ 0; })
+	struct proc_dir_entry *entry;
+#ifdef CONFIG_SMP
+	enum cpuhp_state online_state;
 #endif
 
-int __ref create_proc_profile(void) /* false positive from hotcpu_notifier */
-{
-	struct proc_dir_entry *entry;
 	int err = 0;
 
 	if (!prof_on)
 		return 0;
-
-	cpu_notifier_register_begin();
-
-	if (create_hash_tables()) {
-		err = -ENOMEM;
-		goto out;
-	}
-
+#ifdef CONFIG_SMP
+	err = cpuhp_setup_state(CPUHP_PROFILE_PREPARE, "PROFILE_PREPARE",
+				profile_prepare_cpu, profile_dead_cpu);
+	if (err)
+		return err;
+
+	err = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "AP_PROFILE_ONLINE",
+				profile_online_cpu, NULL);
+	if (err < 0)
+		goto err_state_prep;
+	online_state = err;
+	err = 0;
+#endif
 	entry = proc_create("profile", S_IWUSR | S_IRUGO,
 			    NULL, &proc_profile_operations);
 	if (!entry)
-		goto out;
+		goto err_state_onl;
 	proc_set_size(entry, (1 + prof_len) * sizeof(atomic_t));
-	__hotcpu_notifier(profile_cpu_callback, 0);
 
-out:
-	cpu_notifier_register_done();
+	return err;
+err_state_onl:
+#ifdef CONFIG_SMP
+	cpuhp_remove_state(online_state);
+err_state_prep:
+	cpuhp_remove_state(CPUHP_PROFILE_PREPARE);
+#endif
 	return err;
 }
 subsys_initcall(create_proc_profile);

[patch V2 56/67] timers/core: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0055-timers-core-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 3087 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

When tearing down, call timers_dead_cpu() before notify_dead().
There is a hidden dependency between:

 - timers
 - Block multiqueue
 - rcutree

If timers_dead_cpu() comes later than blk_mq_queue_reinit_notify()
that latter function causes a RCU stall.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |    1 +
 include/linux/timer.h      |    6 ++++++
 kernel/cpu.c               |    5 +++++
 kernel/time/timer.c        |   25 ++-----------------------
 4 files changed, 14 insertions(+), 23 deletions(-)

--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -17,6 +17,7 @@ enum cpuhp_state {
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,
+	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
--- a/include/linux/timer.h
+++ b/include/linux/timer.h
@@ -273,4 +273,10 @@ unsigned long __round_jiffies_up_relativ
 unsigned long round_jiffies_up(unsigned long j);
 unsigned long round_jiffies_up_relative(unsigned long j);
 
+#ifdef CONFIG_HOTPLUG_CPU
+int timers_dead_cpu(unsigned int cpu);
+#else
+#define timers_dead_cpu NULL
+#endif
+
 #endif
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1195,6 +1195,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.startup = hrtimers_prepare_cpu,
 		.teardown = hrtimers_dead_cpu,
 	},
+	[CPUHP_TIMERS_DEAD] = {
+		.name = "timers dead",
+		.startup = NULL,
+		.teardown = timers_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
--- a/kernel/time/timer.c
+++ b/kernel/time/timer.c
@@ -1804,7 +1804,7 @@ static void migrate_timer_list(struct ti
 	}
 }
 
-static void migrate_timers(int cpu)
+int timers_dead_cpu(unsigned int cpu)
 {
 	struct timer_base *old_base;
 	struct timer_base *new_base;
@@ -1831,29 +1831,9 @@ static void migrate_timers(int cpu)
 		spin_unlock_irq(&new_base->lock);
 		put_cpu_ptr(&timer_bases);
 	}
+	return 0;
 }
 
-static int timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
-{
-	switch (action) {
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		migrate_timers((long)hcpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static inline void timer_register_cpu_notifier(void)
-{
-	cpu_notifier(timer_cpu_notify, 0);
-}
-#else
-static inline void timer_register_cpu_notifier(void) { }
 #endif /* CONFIG_HOTPLUG_CPU */
 
 static void __init init_timer_cpu(int cpu)
@@ -1881,7 +1861,6 @@ void __init init_timers(void)
 {
 	init_timer_cpus();
 	init_timer_stats();
-	timer_register_cpu_notifier();
 	open_softirq(TIMER_SOFTIRQ, run_timer_softirq);
 }
 

[tip:smp/hotplug] timers/core: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  24f73b99716a9cd8cbb345c41ced6b3b5ed94006
Gitweb:     http://git.kernel.org/tip/24f73b99716a9cd8cbb345c41ced6b3b5ed94006
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:16:59 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:42 +0200

timers/core: Convert to hotplug state machine

When tearing down, call timers_dead_cpu() before notify_dead().
There is a hidden dependency between:

 - timers
 - block multiqueue
 - rcutree

If timers_dead_cpu() comes later than blk_mq_queue_reinit_notify()
that latter function causes a RCU stall.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.566790058@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  1 +
 include/linux/timer.h      |  6 ++++++
 kernel/cpu.c               |  5 +++++
 kernel/time/timer.c        | 25 ++-----------------------
 4 files changed, 14 insertions(+), 23 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 04ecc55..15d46d2 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -16,6 +16,7 @@ enum cpuhp_state {
 	CPUHP_X86_APB_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_HRTIMERS_PREPARE,
+	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
diff --git a/include/linux/timer.h b/include/linux/timer.h
index 4419506..51d601f 100644
--- a/include/linux/timer.h
+++ b/include/linux/timer.h
@@ -273,4 +273,10 @@ unsigned long __round_jiffies_up_relative(unsigned long j, int cpu);
 unsigned long round_jiffies_up(unsigned long j);
 unsigned long round_jiffies_up_relative(unsigned long j);
 
+#ifdef CONFIG_HOTPLUG_CPU
+int timers_dead_cpu(unsigned int cpu);
+#else
+#define timers_dead_cpu NULL
+#endif
+
 #endif
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 85500e7..e1017d9 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1195,6 +1195,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = hrtimers_prepare_cpu,
 		.teardown = hrtimers_dead_cpu,
 	},
+	[CPUHP_TIMERS_DEAD] = {
+		.name = "timers dead",
+		.startup = NULL,
+		.teardown = timers_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
diff --git a/kernel/time/timer.c b/kernel/time/timer.c
index cb9ab40..555670a 100644
--- a/kernel/time/timer.c
+++ b/kernel/time/timer.c
@@ -1804,7 +1804,7 @@ static void migrate_timer_list(struct timer_base *new_base, struct hlist_head *h
 	}
 }
 
-static void migrate_timers(int cpu)
+int timers_dead_cpu(unsigned int cpu)
 {
 	struct timer_base *old_base;
 	struct timer_base *new_base;
@@ -1831,29 +1831,9 @@ static void migrate_timers(int cpu)
 		spin_unlock_irq(&new_base->lock);
 		put_cpu_ptr(&timer_bases);
 	}
+	return 0;
 }
 
-static int timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
-{
-	switch (action) {
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		migrate_timers((long)hcpu);
-		break;
-	default:
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static inline void timer_register_cpu_notifier(void)
-{
-	cpu_notifier(timer_cpu_notify, 0);
-}
-#else
-static inline void timer_register_cpu_notifier(void) { }
 #endif /* CONFIG_HOTPLUG_CPU */
 
 static void __init init_timer_cpu(int cpu)
@@ -1881,7 +1861,6 @@ void __init init_timers(void)
 {
 	init_timer_cpus();
 	init_timer_stats();
-	timer_register_cpu_notifier();
 	open_softirq(TIMER_SOFTIRQ, run_timer_softirq);
 }
 

[PATCH] timers/core: Correct callback order during CPU hot plug.

[Richard Cochran]

On the tear-down path, the dead CPU callback for the timers was
misplaced within the 'cpuhp_state' enumeration.  There is a hidden
dependency between the timers and block multiqueue.  The timers
callback must happen before the block multiqueue callback otherwise a
RCU stall occurs.

This patch fixes the order within the hot plug state machine.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Fixes: 24f73b99716a ("timers/core: Convert to hotplug state machine")

Cc: John Stultz <john.stultz@linaro.org>
Cc: Jon Hunter <jonathanh@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
---
 include/linux/cpuhotplug.h |  2 +-
 kernel/cpu.c               | 15 ++++++++++-----
 2 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6d405db..242bf53 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,9 +20,9 @@ enum cpuhp_state {
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
-	CPUHP_TIMERS_DEAD,
 	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
+	CPUHP_TIMERS_DEAD,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 67f4943..ec12b72 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1208,11 +1208,6 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = smpcfd_prepare_cpu,
 		.teardown = smpcfd_dead_cpu,
 	},
-	[CPUHP_TIMERS_DEAD] = {
-		.name = "timers dead",
-		.startup = NULL,
-		.teardown = timers_dead_cpu,
-	},
 	[CPUHP_RCUTREE_PREP] = {
 		.name = "RCU-tree prepare",
 		.startup = rcutree_prepare_cpu,
@@ -1229,6 +1224,16 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.skip_onerr		= true,
 		.cant_stop		= true,
 	},
+	/*
+	 * On the tear-down path, timers_dead_cpu() must be invoked
+	 * before blk_mq_queue_reinit_notify() from notify_dead(),
+	 * otherwise a RCU stall occurs.
+	 */
+	[CPUHP_TIMERS_DEAD] = {
+		.name = "timers dead",
+		.startup = NULL,
+		.teardown = timers_dead_cpu,
+	},
 	/* Kicks the plugged cpu into life */
 	[CPUHP_BRINGUP_CPU] = {
 		.name			= "cpu:bringup",
-- 
2.1.4

Re: [PATCH] timers/core: Correct callback order during CPU hot plug.

[Jon Hunter]

Hi Richard,

On 27/07/16 10:08, Richard Cochran wrote:
> On the tear-down path, the dead CPU callback for the timers was
> misplaced within the 'cpuhp_state' enumeration.  There is a hidden
> dependency between the timers and block multiqueue.  The timers
> callback must happen before the block multiqueue callback otherwise a
> RCU stall occurs.
> 
> This patch fixes the order within the hot plug state machine.
> 
> Signed-off-by: Richard Cochran <rcochran@linutronix.de>
> Fixes: 24f73b99716a ("timers/core: Convert to hotplug state machine")
> 
> Cc: John Stultz <john.stultz@linaro.org>
> Cc: Jon Hunter <jonathanh@nvidia.com>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Oleg Nesterov <oleg@redhat.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: rt@linutronix.de
> ---
>  include/linux/cpuhotplug.h |  2 +-
>  kernel/cpu.c               | 15 ++++++++++-----
>  2 files changed, 11 insertions(+), 6 deletions(-)
> 
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index 6d405db..242bf53 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -20,9 +20,9 @@ enum cpuhp_state {
>  	CPUHP_PROFILE_PREPARE,
>  	CPUHP_X2APIC_PREPARE,
>  	CPUHP_SMPCFD_PREPARE,
> -	CPUHP_TIMERS_DEAD,
>  	CPUHP_RCUTREE_PREP,
>  	CPUHP_NOTIFY_PREPARE,
> +	CPUHP_TIMERS_DEAD,
>  	CPUHP_BRINGUP_CPU,
>  	CPUHP_AP_IDLE_DEAD,
>  	CPUHP_AP_OFFLINE,
> diff --git a/kernel/cpu.c b/kernel/cpu.c
> index 67f4943..ec12b72 100644
> --- a/kernel/cpu.c
> +++ b/kernel/cpu.c
> @@ -1208,11 +1208,6 @@ static struct cpuhp_step cpuhp_bp_states[] = {
>  		.startup = smpcfd_prepare_cpu,
>  		.teardown = smpcfd_dead_cpu,
>  	},
> -	[CPUHP_TIMERS_DEAD] = {
> -		.name = "timers dead",
> -		.startup = NULL,
> -		.teardown = timers_dead_cpu,
> -	},
>  	[CPUHP_RCUTREE_PREP] = {
>  		.name = "RCU-tree prepare",
>  		.startup = rcutree_prepare_cpu,
> @@ -1229,6 +1224,16 @@ static struct cpuhp_step cpuhp_bp_states[] = {
>  		.skip_onerr		= true,
>  		.cant_stop		= true,
>  	},
> +	/*
> +	 * On the tear-down path, timers_dead_cpu() must be invoked
> +	 * before blk_mq_queue_reinit_notify() from notify_dead(),
> +	 * otherwise a RCU stall occurs.
> +	 */
> +	[CPUHP_TIMERS_DEAD] = {
> +		.name = "timers dead",
> +		.startup = NULL,
> +		.teardown = timers_dead_cpu,
> +	},
>  	/* Kicks the plugged cpu into life */
>  	[CPUHP_BRINGUP_CPU] = {
>  		.name			= "cpu:bringup",
> 

This fixes the RCU stalls I was seeing during suspend on Tegra and so ...

Tested-by: Jon Hunter <jonathanh@nvidia.com>

Cheers!
Jon

-- 
nvpublic

[tip:smp/hotplug] timers/core: Correct callback order during CPU hot plug

[tip-bot for Richard Cochran]

Commit-ID:  c713c8cb2055f5f3b32ee4315be589177a2658cc
Gitweb:     http://git.kernel.org/tip/c713c8cb2055f5f3b32ee4315be589177a2658cc
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 27 Jul 2016 11:08:18 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 27 Jul 2016 12:38:08 +0200

timers/core: Correct callback order during CPU hot plug

On the tear-down path, the dead CPU callback for the timers was
misplaced within the 'cpuhp_state' enumeration.  There is a hidden
dependency between the timers and block multiqueue.  The timers
callback must happen before the block multiqueue callback otherwise a
RCU stall occurs.

Move the timers callback to the proper place in the state machine.

Reported-and-tested-by: Jon Hunter <jonathanh@nvidia.com>
Reported-by: kbuild test robot <lkp@intel.com>
Fixes: 24f73b99716a ("timers/core: Convert to hotplug state machine")
Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: John Stultz <john.stultz@linaro.org>
Cc: rt@linutronix.de
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1469610498-25914-1-git-send-email-rcochran@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 include/linux/cpuhotplug.h |  2 +-
 kernel/cpu.c               | 15 ++++++++++-----
 2 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6d405db..242bf53 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,9 +20,9 @@ enum cpuhp_state {
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
-	CPUHP_TIMERS_DEAD,
 	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
+	CPUHP_TIMERS_DEAD,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 67f4943..ec12b72 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1208,11 +1208,6 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = smpcfd_prepare_cpu,
 		.teardown = smpcfd_dead_cpu,
 	},
-	[CPUHP_TIMERS_DEAD] = {
-		.name = "timers dead",
-		.startup = NULL,
-		.teardown = timers_dead_cpu,
-	},
 	[CPUHP_RCUTREE_PREP] = {
 		.name = "RCU-tree prepare",
 		.startup = rcutree_prepare_cpu,
@@ -1229,6 +1224,16 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.skip_onerr		= true,
 		.cant_stop		= true,
 	},
+	/*
+	 * On the tear-down path, timers_dead_cpu() must be invoked
+	 * before blk_mq_queue_reinit_notify() from notify_dead(),
+	 * otherwise a RCU stall occurs.
+	 */
+	[CPUHP_TIMERS_DEAD] = {
+		.name = "timers dead",
+		.startup = NULL,
+		.teardown = timers_dead_cpu,
+	},
 	/* Kicks the plugged cpu into life */
 	[CPUHP_BRINGUP_CPU] = {
 		.name			= "cpu:bringup",

[tip:smp/hotplug] timers/core: Correct callback order during CPU hot plug

[tip-bot for Richard Cochran]

Commit-ID:  4fae16dffb812f0e0d98a0b2b0856ca48ca63e6c
Gitweb:     http://git.kernel.org/tip/4fae16dffb812f0e0d98a0b2b0856ca48ca63e6c
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 27 Jul 2016 11:08:18 +0200
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 28 Jul 2016 18:56:22 +0200

timers/core: Correct callback order during CPU hot plug

On the tear-down path, the dead CPU callback for the timers was
misplaced within the 'cpuhp_state' enumeration.  There is a hidden
dependency between the timers and block multiqueue.  The timers
callback must happen before the block multiqueue callback otherwise a
RCU stall occurs.

Move the timers callback to the proper place in the state machine.

Reported-and-tested-by: Jon Hunter <jonathanh@nvidia.com>
Reported-by: kbuild test robot <lkp@intel.com>
Fixes: 24f73b99716a ("timers/core: Convert to hotplug state machine")
Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: John Stultz <john.stultz@linaro.org>
Cc: rt@linutronix.de
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1469610498-25914-1-git-send-email-rcochran@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  2 +-
 kernel/cpu.c               | 15 ++++++++++-----
 2 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 6d405db..242bf53 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,9 +20,9 @@ enum cpuhp_state {
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
-	CPUHP_TIMERS_DEAD,
 	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
+	CPUHP_TIMERS_DEAD,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index f24f459..341bf80 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1200,11 +1200,6 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = smpcfd_prepare_cpu,
 		.teardown = smpcfd_dead_cpu,
 	},
-	[CPUHP_TIMERS_DEAD] = {
-		.name = "timers dead",
-		.startup = NULL,
-		.teardown = timers_dead_cpu,
-	},
 	[CPUHP_RCUTREE_PREP] = {
 		.name = "RCU-tree prepare",
 		.startup = rcutree_prepare_cpu,
@@ -1221,6 +1216,16 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.skip_onerr		= true,
 		.cant_stop		= true,
 	},
+	/*
+	 * On the tear-down path, timers_dead_cpu() must be invoked
+	 * before blk_mq_queue_reinit_notify() from notify_dead(),
+	 * otherwise a RCU stall occurs.
+	 */
+	[CPUHP_TIMERS_DEAD] = {
+		.name = "timers dead",
+		.startup = NULL,
+		.teardown = timers_dead_cpu,
+	},
 	/* Kicks the plugged cpu into life */
 	[CPUHP_BRINGUP_CPU] = {
 		.name			= "cpu:bringup",

[patch V2 58/67] x86/x2apic: Convert to CPU hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0057-x86-x2apic-Convert-to-CPU-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 4079 bytes --]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Len Brown <len.brown@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mathias Krause <minipli@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/x86/kernel/apic/x2apic_cluster.c | 80 +++++++++++++----------------------
 include/linux/cpuhotplug.h            |  1 +
 2 files changed, 31 insertions(+), 50 deletions(-)

diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index aca8b75..b5da5a8 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -152,68 +152,48 @@ static void init_x2apic_ldr(void)
 	}
 }
 
- /*
-  * At CPU state changes, update the x2apic cluster sibling info.
-  */
-static int
-update_clusterinfo(struct notifier_block *nfb, unsigned long action, void *hcpu)
+/*
+ * At CPU state changes, update the x2apic cluster sibling info.
+ */
+int x2apic_prepare_cpu(unsigned int cpu)
 {
-	unsigned int this_cpu = (unsigned long)hcpu;
-	unsigned int cpu;
-	int err = 0;
-
-	switch (action) {
-	case CPU_UP_PREPARE:
-		if (!zalloc_cpumask_var(&per_cpu(cpus_in_cluster, this_cpu),
-					GFP_KERNEL)) {
-			err = -ENOMEM;
-		} else if (!zalloc_cpumask_var(&per_cpu(ipi_mask, this_cpu),
-					       GFP_KERNEL)) {
-			free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
-			err = -ENOMEM;
-		}
-		break;
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-	case CPU_DEAD:
-		for_each_online_cpu(cpu) {
-			if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
-				continue;
-			cpumask_clear_cpu(this_cpu, per_cpu(cpus_in_cluster, cpu));
-			cpumask_clear_cpu(cpu, per_cpu(cpus_in_cluster, this_cpu));
-		}
-		free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
-		free_cpumask_var(per_cpu(ipi_mask, this_cpu));
-		break;
+	if (!zalloc_cpumask_var(&per_cpu(cpus_in_cluster, cpu), GFP_KERNEL))
+		return -ENOMEM;
+
+	if (!zalloc_cpumask_var(&per_cpu(ipi_mask, cpu), GFP_KERNEL)) {
+		free_cpumask_var(per_cpu(cpus_in_cluster, cpu));
+		return -ENOMEM;
 	}
 
-	return notifier_from_errno(err);
+	return 0;
 }
 
-static struct notifier_block x2apic_cpu_notifier = {
-	.notifier_call = update_clusterinfo,
-};
-
-static int x2apic_init_cpu_notifier(void)
+int x2apic_dead_cpu(unsigned int this_cpu)
 {
-	int cpu = smp_processor_id();
-
-	zalloc_cpumask_var(&per_cpu(cpus_in_cluster, cpu), GFP_KERNEL);
-	zalloc_cpumask_var(&per_cpu(ipi_mask, cpu), GFP_KERNEL);
+	int cpu;
 
-	BUG_ON(!per_cpu(cpus_in_cluster, cpu) || !per_cpu(ipi_mask, cpu));
-
-	cpumask_set_cpu(cpu, per_cpu(cpus_in_cluster, cpu));
-	register_hotcpu_notifier(&x2apic_cpu_notifier);
-	return 1;
+	for_each_online_cpu(cpu) {
+		if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
+			continue;
+		cpumask_clear_cpu(this_cpu, per_cpu(cpus_in_cluster, cpu));
+		cpumask_clear_cpu(cpu, per_cpu(cpus_in_cluster, this_cpu));
+	}
+	free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
+	free_cpumask_var(per_cpu(ipi_mask, this_cpu));
+	return 0;
 }
 
 static int x2apic_cluster_probe(void)
 {
-	if (x2apic_mode)
-		return x2apic_init_cpu_notifier();
-	else
+	int cpu = smp_processor_id();
+
+	if (!x2apic_mode)
 		return 0;
+
+	cpumask_set_cpu(cpu, per_cpu(cpus_in_cluster, cpu));
+	cpuhp_setup_state(CPUHP_X2APIC_PREPARE, "X2APIC_PREPARE",
+			  x2apic_prepare_cpu, x2apic_dead_cpu);
+	return 1;
 }
 
 static const struct cpumask *x2apic_cluster_target_cpus(void)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index a38fb6f..c631cd4 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -18,6 +18,7 @@ enum cpuhp_state {
 	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
+	CPUHP_X2APIC_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
-- 
2.8.1

[tip:smp/hotplug] x86/x2apic: Convert to CPU hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  6b2c28471de550308784560206c3365e5179d42f
Gitweb:     http://git.kernel.org/tip/6b2c28471de550308784560206c3365e5179d42f
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:00 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:42 +0200

x86/x2apic: Convert to CPU hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Len Brown <len.brown@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mathias Krause <minipli@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.736898691@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apic/x2apic_cluster.c | 80 +++++++++++++----------------------
 include/linux/cpuhotplug.h            |  1 +
 2 files changed, 31 insertions(+), 50 deletions(-)

diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index aca8b75..b5da5a8 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -152,68 +152,48 @@ static void init_x2apic_ldr(void)
 	}
 }
 
- /*
-  * At CPU state changes, update the x2apic cluster sibling info.
-  */
-static int
-update_clusterinfo(struct notifier_block *nfb, unsigned long action, void *hcpu)
+/*
+ * At CPU state changes, update the x2apic cluster sibling info.
+ */
+int x2apic_prepare_cpu(unsigned int cpu)
 {
-	unsigned int this_cpu = (unsigned long)hcpu;
-	unsigned int cpu;
-	int err = 0;
-
-	switch (action) {
-	case CPU_UP_PREPARE:
-		if (!zalloc_cpumask_var(&per_cpu(cpus_in_cluster, this_cpu),
-					GFP_KERNEL)) {
-			err = -ENOMEM;
-		} else if (!zalloc_cpumask_var(&per_cpu(ipi_mask, this_cpu),
-					       GFP_KERNEL)) {
-			free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
-			err = -ENOMEM;
-		}
-		break;
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-	case CPU_DEAD:
-		for_each_online_cpu(cpu) {
-			if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
-				continue;
-			cpumask_clear_cpu(this_cpu, per_cpu(cpus_in_cluster, cpu));
-			cpumask_clear_cpu(cpu, per_cpu(cpus_in_cluster, this_cpu));
-		}
-		free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
-		free_cpumask_var(per_cpu(ipi_mask, this_cpu));
-		break;
+	if (!zalloc_cpumask_var(&per_cpu(cpus_in_cluster, cpu), GFP_KERNEL))
+		return -ENOMEM;
+
+	if (!zalloc_cpumask_var(&per_cpu(ipi_mask, cpu), GFP_KERNEL)) {
+		free_cpumask_var(per_cpu(cpus_in_cluster, cpu));
+		return -ENOMEM;
 	}
 
-	return notifier_from_errno(err);
+	return 0;
 }
 
-static struct notifier_block x2apic_cpu_notifier = {
-	.notifier_call = update_clusterinfo,
-};
-
-static int x2apic_init_cpu_notifier(void)
+int x2apic_dead_cpu(unsigned int this_cpu)
 {
-	int cpu = smp_processor_id();
-
-	zalloc_cpumask_var(&per_cpu(cpus_in_cluster, cpu), GFP_KERNEL);
-	zalloc_cpumask_var(&per_cpu(ipi_mask, cpu), GFP_KERNEL);
+	int cpu;
 
-	BUG_ON(!per_cpu(cpus_in_cluster, cpu) || !per_cpu(ipi_mask, cpu));
-
-	cpumask_set_cpu(cpu, per_cpu(cpus_in_cluster, cpu));
-	register_hotcpu_notifier(&x2apic_cpu_notifier);
-	return 1;
+	for_each_online_cpu(cpu) {
+		if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
+			continue;
+		cpumask_clear_cpu(this_cpu, per_cpu(cpus_in_cluster, cpu));
+		cpumask_clear_cpu(cpu, per_cpu(cpus_in_cluster, this_cpu));
+	}
+	free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
+	free_cpumask_var(per_cpu(ipi_mask, this_cpu));
+	return 0;
 }
 
 static int x2apic_cluster_probe(void)
 {
-	if (x2apic_mode)
-		return x2apic_init_cpu_notifier();
-	else
+	int cpu = smp_processor_id();
+
+	if (!x2apic_mode)
 		return 0;
+
+	cpumask_set_cpu(cpu, per_cpu(cpus_in_cluster, cpu));
+	cpuhp_setup_state(CPUHP_X2APIC_PREPARE, "X2APIC_PREPARE",
+			  x2apic_prepare_cpu, x2apic_dead_cpu);
+	return 1;
 }
 
 static const struct cpumask *x2apic_cluster_target_cpus(void)
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ace5ad0..7817082 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -17,6 +17,7 @@ enum cpuhp_state {
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
+	CPUHP_X2APIC_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,

[patch V2 59/67] smp: Convert core to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0058-smp-Convert-core-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 5424 bytes --]

From: Richard Weinberger <richard@nod.at>

Install the callbacks via the state machine. They are installed at runtime so
smpcfd_prepare_cpu() needs to be invoked by the boot-CPU.

Signed-off-by: Richard Weinberger <richard@nod.at>
[ Added the dropped CPU dying case back in. ]
Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Davidlohr Bueso <dave@stgolabs>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |    2 +
 include/linux/smp.h        |    5 ++
 kernel/cpu.c               |    9 +++++
 kernel/smp.c               |   79 ++++++++++++++++++---------------------------
 4 files changed, 48 insertions(+), 47 deletions(-)

--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -19,6 +19,7 @@ enum cpuhp_state {
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
+	CPUHP_SMPCFD_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
@@ -58,6 +59,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
+	CPUHP_AP_SMPCFD_DYING,
 	CPUHP_AP_X86_TBOOT_DYING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
--- a/include/linux/smp.h
+++ b/include/linux/smp.h
@@ -196,4 +196,9 @@ extern void arch_enable_nonboot_cpus_end
 
 void smp_setup_processor_id(void);
 
+/* SMP core functions */
+int smpcfd_prepare_cpu(unsigned int cpu);
+int smpcfd_dead_cpu(unsigned int cpu);
+int smpcfd_dying_cpu(unsigned int cpu);
+
 #endif /* __LINUX_SMP_H */
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1195,6 +1195,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.startup = hrtimers_prepare_cpu,
 		.teardown = hrtimers_dead_cpu,
 	},
+	[CPUHP_SMPCFD_PREPARE] = {
+		.name = "SMPCFD prepare",
+		.startup = smpcfd_prepare_cpu,
+		.teardown = smpcfd_dead_cpu,
+	},
 	[CPUHP_TIMERS_DEAD] = {
 		.name = "timers dead",
 		.startup = NULL,
@@ -1218,6 +1223,10 @@ static struct cpuhp_step cpuhp_bp_states
 		.teardown		= NULL,
 		.cant_stop		= true,
 	},
+	[CPUHP_AP_SMPCFD_DYING] = {
+		.startup = NULL,
+		.teardown = smpcfd_dying_cpu,
+	},
 	/*
 	 * Handled on controll processor until the plugged processor manages
 	 * this itself.
--- a/kernel/smp.c
+++ b/kernel/smp.c
@@ -33,69 +33,54 @@ static DEFINE_PER_CPU_SHARED_ALIGNED(str
 
 static void flush_smp_call_function_queue(bool warn_cpu_offline);
 
-static int
-hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu)
+int smpcfd_prepare_cpu(unsigned int cpu)
 {
-	long cpu = (long)hcpu;
 	struct call_function_data *cfd = &per_cpu(cfd_data, cpu);
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL,
-				cpu_to_node(cpu)))
-			return notifier_from_errno(-ENOMEM);
-		cfd->csd = alloc_percpu(struct call_single_data);
-		if (!cfd->csd) {
-			free_cpumask_var(cfd->cpumask);
-			return notifier_from_errno(-ENOMEM);
-		}
-		break;
-
-#ifdef CONFIG_HOTPLUG_CPU
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		/* Fall-through to the CPU_DEAD[_FROZEN] case. */
-
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
+	if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL,
+				     cpu_to_node(cpu)))
+		return -ENOMEM;
+	cfd->csd = alloc_percpu(struct call_single_data);
+	if (!cfd->csd) {
 		free_cpumask_var(cfd->cpumask);
-		free_percpu(cfd->csd);
-		break;
+		return -ENOMEM;
+	}
 
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		/*
-		 * The IPIs for the smp-call-function callbacks queued by other
-		 * CPUs might arrive late, either due to hardware latencies or
-		 * because this CPU disabled interrupts (inside stop-machine)
-		 * before the IPIs were sent. So flush out any pending callbacks
-		 * explicitly (without waiting for the IPIs to arrive), to
-		 * ensure that the outgoing CPU doesn't go offline with work
-		 * still pending.
-		 */
-		flush_smp_call_function_queue(false);
-		break;
-#endif
-	};
+	return 0;
+}
 
-	return NOTIFY_OK;
+int smpcfd_dead_cpu(unsigned int cpu)
+{
+	struct call_function_data *cfd = &per_cpu(cfd_data, cpu);
+
+	free_cpumask_var(cfd->cpumask);
+	free_percpu(cfd->csd);
+	return 0;
 }
 
-static struct notifier_block hotplug_cfd_notifier = {
-	.notifier_call		= hotplug_cfd,
-};
+int smpcfd_dying_cpu(unsigned int cpu)
+{
+	/*
+	 * The IPIs for the smp-call-function callbacks queued by other
+	 * CPUs might arrive late, either due to hardware latencies or
+	 * because this CPU disabled interrupts (inside stop-machine)
+	 * before the IPIs were sent. So flush out any pending callbacks
+	 * explicitly (without waiting for the IPIs to arrive), to
+	 * ensure that the outgoing CPU doesn't go offline with work
+	 * still pending.
+	 */
+	flush_smp_call_function_queue(false);
+	return 0;
+}
 
 void __init call_function_init(void)
 {
-	void *cpu = (void *)(long)smp_processor_id();
 	int i;
 
 	for_each_possible_cpu(i)
 		init_llist_head(&per_cpu(call_single_queue, i));
 
-	hotplug_cfd(&hotplug_cfd_notifier, CPU_UP_PREPARE, cpu);
-	register_cpu_notifier(&hotplug_cfd_notifier);
+	smpcfd_prepare_cpu(smp_processor_id());
 }
 
 /*

[tip:smp/hotplug] smp/cfd: Convert core to hotplug state machine

[tip-bot for Richard Weinberger]

Commit-ID:  31487f8328f20fdb302430b020a5d6e8446c1971
Gitweb:     http://git.kernel.org/tip/31487f8328f20fdb302430b020a5d6e8446c1971
Author:     Richard Weinberger <richard@nod.at>
AuthorDate: Wed, 13 Jul 2016 17:17:01 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:43 +0200

smp/cfd: Convert core to hotplug state machine

Install the callbacks via the state machine. They are installed at runtime so
smpcfd_prepare_cpu() needs to be invoked by the boot-CPU.

Signed-off-by: Richard Weinberger <richard@nod.at>
[ Added the dropped CPU dying case back in. ]
Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Davidlohr Bueso <dave@stgolabs>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.818376366@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |  2 ++
 include/linux/smp.h        |  5 +++
 kernel/cpu.c               |  9 ++++++
 kernel/smp.c               | 79 +++++++++++++++++++---------------------------
 4 files changed, 48 insertions(+), 47 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 7817082..b5cf01a 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -18,6 +18,7 @@ enum cpuhp_state {
 	CPUHP_HRTIMERS_PREPARE,
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
+	CPUHP_SMPCFD_PREPARE,
 	CPUHP_TIMERS_DEAD,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
@@ -57,6 +58,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_CORESIGHT4_STARTING,
 	CPUHP_AP_ARM64_ISNDEP_STARTING,
 	CPUHP_AP_LEDTRIG_STARTING,
+	CPUHP_AP_SMPCFD_DYING,
 	CPUHP_AP_X86_TBOOT_DYING,
 	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
diff --git a/include/linux/smp.h b/include/linux/smp.h
index c441407..eccae469 100644
--- a/include/linux/smp.h
+++ b/include/linux/smp.h
@@ -196,4 +196,9 @@ extern void arch_enable_nonboot_cpus_end(void);
 
 void smp_setup_processor_id(void);
 
+/* SMP core functions */
+int smpcfd_prepare_cpu(unsigned int cpu);
+int smpcfd_dead_cpu(unsigned int cpu);
+int smpcfd_dying_cpu(unsigned int cpu);
+
 #endif /* __LINUX_SMP_H */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index e1017d9..008e2fd 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1195,6 +1195,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = hrtimers_prepare_cpu,
 		.teardown = hrtimers_dead_cpu,
 	},
+	[CPUHP_SMPCFD_PREPARE] = {
+		.name = "SMPCFD prepare",
+		.startup = smpcfd_prepare_cpu,
+		.teardown = smpcfd_dead_cpu,
+	},
 	[CPUHP_TIMERS_DEAD] = {
 		.name = "timers dead",
 		.startup = NULL,
@@ -1218,6 +1223,10 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.teardown		= NULL,
 		.cant_stop		= true,
 	},
+	[CPUHP_AP_SMPCFD_DYING] = {
+		.startup = NULL,
+		.teardown = smpcfd_dying_cpu,
+	},
 	/*
 	 * Handled on controll processor until the plugged processor manages
 	 * this itself.
diff --git a/kernel/smp.c b/kernel/smp.c
index 7416544..7180491 100644
--- a/kernel/smp.c
+++ b/kernel/smp.c
@@ -33,69 +33,54 @@ static DEFINE_PER_CPU_SHARED_ALIGNED(struct llist_head, call_single_queue);
 
 static void flush_smp_call_function_queue(bool warn_cpu_offline);
 
-static int
-hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu)
+int smpcfd_prepare_cpu(unsigned int cpu)
 {
-	long cpu = (long)hcpu;
 	struct call_function_data *cfd = &per_cpu(cfd_data, cpu);
 
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL,
-				cpu_to_node(cpu)))
-			return notifier_from_errno(-ENOMEM);
-		cfd->csd = alloc_percpu(struct call_single_data);
-		if (!cfd->csd) {
-			free_cpumask_var(cfd->cpumask);
-			return notifier_from_errno(-ENOMEM);
-		}
-		break;
-
-#ifdef CONFIG_HOTPLUG_CPU
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		/* Fall-through to the CPU_DEAD[_FROZEN] case. */
-
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
+	if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL,
+				     cpu_to_node(cpu)))
+		return -ENOMEM;
+	cfd->csd = alloc_percpu(struct call_single_data);
+	if (!cfd->csd) {
 		free_cpumask_var(cfd->cpumask);
-		free_percpu(cfd->csd);
-		break;
+		return -ENOMEM;
+	}
 
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		/*
-		 * The IPIs for the smp-call-function callbacks queued by other
-		 * CPUs might arrive late, either due to hardware latencies or
-		 * because this CPU disabled interrupts (inside stop-machine)
-		 * before the IPIs were sent. So flush out any pending callbacks
-		 * explicitly (without waiting for the IPIs to arrive), to
-		 * ensure that the outgoing CPU doesn't go offline with work
-		 * still pending.
-		 */
-		flush_smp_call_function_queue(false);
-		break;
-#endif
-	};
+	return 0;
+}
+
+int smpcfd_dead_cpu(unsigned int cpu)
+{
+	struct call_function_data *cfd = &per_cpu(cfd_data, cpu);
 
-	return NOTIFY_OK;
+	free_cpumask_var(cfd->cpumask);
+	free_percpu(cfd->csd);
+	return 0;
 }
 
-static struct notifier_block hotplug_cfd_notifier = {
-	.notifier_call		= hotplug_cfd,
-};
+int smpcfd_dying_cpu(unsigned int cpu)
+{
+	/*
+	 * The IPIs for the smp-call-function callbacks queued by other
+	 * CPUs might arrive late, either due to hardware latencies or
+	 * because this CPU disabled interrupts (inside stop-machine)
+	 * before the IPIs were sent. So flush out any pending callbacks
+	 * explicitly (without waiting for the IPIs to arrive), to
+	 * ensure that the outgoing CPU doesn't go offline with work
+	 * still pending.
+	 */
+	flush_smp_call_function_queue(false);
+	return 0;
+}
 
 void __init call_function_init(void)
 {
-	void *cpu = (void *)(long)smp_processor_id();
 	int i;
 
 	for_each_possible_cpu(i)
 		init_llist_head(&per_cpu(call_single_queue, i));
 
-	hotplug_cfd(&hotplug_cfd_notifier, CPU_UP_PREPARE, cpu);
-	register_cpu_notifier(&hotplug_cfd_notifier);
+	smpcfd_prepare_cpu(smp_processor_id());
 }
 
 /*

[patch V2 60/67] KVM/arm/arm64/vgic-new: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0059-KVM-arm-arm64-vgic-new-Convert-to-hotplug-state-mach.patch --]
[-- Type: text/plain, Size: 2881 bytes --]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andre Przywara <andre.przywara@arm.com>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Eric Auger <eric.auger@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
---
 include/linux/cpuhotplug.h    |  1 +
 virt/kvm/arm/vgic/vgic-init.c | 31 +++++++++----------------------
 2 files changed, 10 insertions(+), 22 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 078bd22..822df54 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -52,6 +52,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
diff --git a/virt/kvm/arm/vgic/vgic-init.c b/virt/kvm/arm/vgic/vgic-init.c
index a1442f7..2c7f0d5 100644
--- a/virt/kvm/arm/vgic/vgic-init.c
+++ b/virt/kvm/arm/vgic/vgic-init.c
@@ -353,32 +353,19 @@ out:
 
 /* GENERIC PROBE */
 
-static void vgic_init_maintenance_interrupt(void *info)
+static int vgic_init_cpu_starting(unsigned int cpu)
 {
 	enable_percpu_irq(kvm_vgic_global_state.maint_irq, 0);
+	return 0;
 }
 
-static int vgic_cpu_notify(struct notifier_block *self,
-			   unsigned long action, void *cpu)
-{
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		vgic_init_maintenance_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(kvm_vgic_global_state.maint_irq);
-		break;
-	}
 
-	return NOTIFY_OK;
+static int vgic_init_cpu_dying(unsigned int cpu)
+{
+	disable_percpu_irq(kvm_vgic_global_state.maint_irq);
+	return 0;
 }
 
-static struct notifier_block vgic_cpu_nb = {
-	.notifier_call = vgic_cpu_notify,
-};
-
 static irqreturn_t vgic_maintenance_handler(int irq, void *data)
 {
 	/*
@@ -434,14 +421,14 @@ int kvm_vgic_hyp_init(void)
 		return ret;
 	}
 
-	ret = __register_cpu_notifier(&vgic_cpu_nb);
+	ret = cpuhp_setup_state(CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
+				"AP_KVM_ARM_VGIC_INIT_STARTING",
+				vgic_init_cpu_starting, vgic_init_cpu_dying);
 	if (ret) {
 		kvm_err("Cannot register vgic CPU notifier\n");
 		goto out_free_irq;
 	}
 
-	on_each_cpu(vgic_init_maintenance_interrupt, NULL, 1);
-
 	kvm_info("vgic interrupt IRQ%d\n", kvm_vgic_global_state.maint_irq);
 	return 0;
 
-- 
2.8.1

Re: [patch V2 60/67] KVM/arm/arm64/vgic-new: Convert to hotplug state machine

[Marc Zyngier]

On 13/07/16 18:17, Anna-Maria Gleixner wrote:
> From: Anna-Maria Gleixner <anna-maria@linutronix.de>
> 
> Install the callbacks via the state machine and let the core invoke
> the callbacks on the already online CPUs.
> 
> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
> Cc: Andre Przywara <andre.przywara@arm.com>
> Cc: Christoffer Dall <christoffer.dall@linaro.org>
> Cc: Eric Auger <eric.auger@linaro.org>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Marc Zyngier <marc.zyngier@arm.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Radim Krcmar <rkrcmar@redhat.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: kvm@vger.kernel.org
> Cc: kvmarm@lists.cs.columbia.edu
> Cc: linux-arm-kernel@lists.infradead.org

For patches 3, 4, 5, 6, 7, 44, 45 and 60:

Acked-by: Marc Zyngier <marc.zyngier@arm.com>

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...

[tip:smp/hotplug] KVM/arm/arm64/vgic-new: Convert to hotplug state machine

[tip-bot for Anna-Maria Gleixner]

Commit-ID:  15d7e3d349a329ad48a29c3a818eacd1c3f7e3ef
Gitweb:     http://git.kernel.org/tip/15d7e3d349a329ad48a29c3a818eacd1c3f7e3ef
Author:     Anna-Maria Gleixner <anna-maria@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:02 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:43 +0200

KVM/arm/arm64/vgic-new: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Andre Przywara <andre.przywara@arm.com>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Eric Auger <eric.auger@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kvm@vger.kernel.org
Cc: kvmarm@lists.cs.columbia.edu
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.900484868@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h    |  1 +
 virt/kvm/arm/vgic/vgic-init.c | 31 +++++++++----------------------
 2 files changed, 10 insertions(+), 22 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index b5cf01a..544b556 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -51,6 +51,7 @@ enum cpuhp_state {
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
+	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,
 	CPUHP_AP_KVM_ARM_TIMER_STARTING,
 	CPUHP_AP_ARM_XEN_STARTING,
diff --git a/virt/kvm/arm/vgic/vgic-init.c b/virt/kvm/arm/vgic/vgic-init.c
index a1442f7..2c7f0d5 100644
--- a/virt/kvm/arm/vgic/vgic-init.c
+++ b/virt/kvm/arm/vgic/vgic-init.c
@@ -353,32 +353,19 @@ out:
 
 /* GENERIC PROBE */
 
-static void vgic_init_maintenance_interrupt(void *info)
+static int vgic_init_cpu_starting(unsigned int cpu)
 {
 	enable_percpu_irq(kvm_vgic_global_state.maint_irq, 0);
+	return 0;
 }
 
-static int vgic_cpu_notify(struct notifier_block *self,
-			   unsigned long action, void *cpu)
-{
-	switch (action) {
-	case CPU_STARTING:
-	case CPU_STARTING_FROZEN:
-		vgic_init_maintenance_interrupt(NULL);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		disable_percpu_irq(kvm_vgic_global_state.maint_irq);
-		break;
-	}
 
-	return NOTIFY_OK;
+static int vgic_init_cpu_dying(unsigned int cpu)
+{
+	disable_percpu_irq(kvm_vgic_global_state.maint_irq);
+	return 0;
 }
 
-static struct notifier_block vgic_cpu_nb = {
-	.notifier_call = vgic_cpu_notify,
-};
-
 static irqreturn_t vgic_maintenance_handler(int irq, void *data)
 {
 	/*
@@ -434,14 +421,14 @@ int kvm_vgic_hyp_init(void)
 		return ret;
 	}
 
-	ret = __register_cpu_notifier(&vgic_cpu_nb);
+	ret = cpuhp_setup_state(CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
+				"AP_KVM_ARM_VGIC_INIT_STARTING",
+				vgic_init_cpu_starting, vgic_init_cpu_dying);
 	if (ret) {
 		kvm_err("Cannot register vgic CPU notifier\n");
 		goto out_free_irq;
 	}
 
-	on_each_cpu(vgic_init_maintenance_interrupt, NULL, 1);
-
 	kvm_info("vgic interrupt IRQ%d\n", kvm_vgic_global_state.maint_irq);
 	return 0;
 

[patch V2 61/67] rcu: Convert rcutree to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0062-rcu-Convert-rcutree-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 7105 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

Straight forward conversion to the state machine. Though the question arises
whether this needs really all these state transitions to work.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 include/linux/cpuhotplug.h |    3 +
 include/linux/rcutiny.h    |    7 +++
 include/linux/rcutree.h    |    7 +++
 kernel/cpu.c               |   14 ++++++
 kernel/rcu/tree.c          |  105 ++++++++++++++++++++++-----------------------
 5 files changed, 83 insertions(+), 53 deletions(-)

--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -21,11 +21,13 @@ enum cpuhp_state {
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
 	CPUHP_TIMERS_DEAD,
+	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
+	CPUHP_AP_RCUTREE_DYING,
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_IRQ_HIP04_STARTING,
@@ -81,6 +83,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_PERF_ARM_CCN_ONLINE,
 	CPUHP_AP_WORKQUEUE_ONLINE,
+	CPUHP_AP_RCUTREE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
--- a/include/linux/rcutiny.h
+++ b/include/linux/rcutiny.h
@@ -243,4 +243,11 @@ static inline void rcu_all_qs(void)
 	barrier(); /* Avoid RCU read-side critical sections leaking across. */
 }
 
+/* RCUtree hotplug events */
+#define rcutree_prepare_cpu      NULL
+#define rcutree_online_cpu       NULL
+#define rcutree_offline_cpu      NULL
+#define rcutree_dead_cpu         NULL
+#define rcutree_dying_cpu        NULL
+
 #endif /* __LINUX_RCUTINY_H */
--- a/include/linux/rcutree.h
+++ b/include/linux/rcutree.h
@@ -111,4 +111,11 @@ bool rcu_is_watching(void);
 
 void rcu_all_qs(void);
 
+/* RCUtree hotplug events */
+int rcutree_prepare_cpu(unsigned int cpu);
+int rcutree_online_cpu(unsigned int cpu);
+int rcutree_offline_cpu(unsigned int cpu);
+int rcutree_dead_cpu(unsigned int cpu);
+int rcutree_dying_cpu(unsigned int cpu);
+
 #endif /* __LINUX_RCUTREE_H */
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1205,6 +1205,11 @@ static struct cpuhp_step cpuhp_bp_states
 		.startup = NULL,
 		.teardown = timers_dead_cpu,
 	},
+	[CPUHP_RCUTREE_PREP] = {
+		.name = "RCU-tree prepare",
+		.startup = rcutree_prepare_cpu,
+		.teardown = rcutree_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1261,6 +1266,10 @@ static struct cpuhp_step cpuhp_ap_states
 		.startup		= sched_cpu_starting,
 		.teardown		= sched_cpu_dying,
 	},
+	[CPUHP_AP_RCUTREE_DYING] = {
+		.startup = NULL,
+		.teardown = rcutree_dying_cpu,
+	},
 	/*
 	 * Low level startup/teardown notifiers. Run with interrupts
 	 * disabled. Will be removed once the notifiers are converted to
@@ -1294,6 +1303,11 @@ static struct cpuhp_step cpuhp_ap_states
 		.startup = workqueue_online_cpu,
 		.teardown = workqueue_offline_cpu,
 	},
+	[CPUHP_AP_RCUTREE_ONLINE] = {
+		.name = "RCU-tree online",
+		.startup = rcutree_online_cpu,
+		.teardown = rcutree_offline_cpu,
+	},
 
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -1073,11 +1073,11 @@ EXPORT_SYMBOL_GPL(rcu_is_watching);
  * offline to continue to use RCU for one jiffy after marking itself
  * offline in the cpu_online_mask.  This leniency is necessary given the
  * non-atomic nature of the online and offline processing, for example,
- * the fact that a CPU enters the scheduler after completing the CPU_DYING
- * notifiers.
+ * the fact that a CPU enters the scheduler after completing the teardown
+ * of the CPU.
  *
- * This is also why RCU internally marks CPUs online during the
- * CPU_UP_PREPARE phase and offline during the CPU_DEAD phase.
+ * This is also why RCU internally marks CPUs online during in the
+ * preparation phase and offline after the CPU has been taken down.
  *
  * Disable checking if in an NMI handler because we cannot safely report
  * errors from NMI handlers anyway.
@@ -3806,12 +3806,58 @@ rcu_init_percpu_data(int cpu, struct rcu
 	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 }
 
-static void rcu_prepare_cpu(int cpu)
+int rcutree_prepare_cpu(unsigned int cpu)
 {
 	struct rcu_state *rsp;
 
 	for_each_rcu_flavor(rsp)
 		rcu_init_percpu_data(cpu, rsp);
+
+	rcu_prepare_kthreads(cpu);
+	rcu_spawn_all_nocb_kthreads(cpu);
+
+	return 0;
+}
+
+static void rcutree_affinity_setting(unsigned int cpu, int outgoing)
+{
+	struct rcu_data *rdp = per_cpu_ptr(rcu_state_p->rda, cpu);
+
+	rcu_boost_kthread_setaffinity(rdp->mynode, outgoing);
+}
+
+int rcutree_online_cpu(unsigned int cpu)
+{
+	sync_sched_exp_online_cleanup(cpu);
+	rcutree_affinity_setting(cpu, -1);
+	return 0;
+}
+
+int rcutree_offline_cpu(unsigned int cpu)
+{
+	rcutree_affinity_setting(cpu, cpu);
+	return 0;
+}
+
+
+int rcutree_dying_cpu(unsigned int cpu)
+{
+	struct rcu_state *rsp;
+
+	for_each_rcu_flavor(rsp)
+		rcu_cleanup_dying_cpu(rsp);
+	return 0;
+}
+
+int rcutree_dead_cpu(unsigned int cpu)
+{
+	struct rcu_state *rsp;
+
+	for_each_rcu_flavor(rsp) {
+		rcu_cleanup_dead_cpu(cpu, rsp);
+		do_nocb_deferred_wakeup(per_cpu_ptr(rsp->rda, cpu));
+	}
+	return 0;
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -3851,52 +3897,6 @@ void rcu_report_dead(unsigned int cpu)
 }
 #endif
 
-/*
- * Handle CPU online/offline notification events.
- */
-int rcu_cpu_notify(struct notifier_block *self,
-		   unsigned long action, void *hcpu)
-{
-	long cpu = (long)hcpu;
-	struct rcu_data *rdp = per_cpu_ptr(rcu_state_p->rda, cpu);
-	struct rcu_node *rnp = rdp->mynode;
-	struct rcu_state *rsp;
-
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		rcu_prepare_cpu(cpu);
-		rcu_prepare_kthreads(cpu);
-		rcu_spawn_all_nocb_kthreads(cpu);
-		break;
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		sync_sched_exp_online_cleanup(cpu);
-		rcu_boost_kthread_setaffinity(rnp, -1);
-		break;
-	case CPU_DOWN_PREPARE:
-		rcu_boost_kthread_setaffinity(rnp, cpu);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		for_each_rcu_flavor(rsp)
-			rcu_cleanup_dying_cpu(rsp);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		for_each_rcu_flavor(rsp) {
-			rcu_cleanup_dead_cpu(cpu, rsp);
-			do_nocb_deferred_wakeup(per_cpu_ptr(rsp->rda, cpu));
-		}
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_OK;
-}
-
 static int rcu_pm_notify(struct notifier_block *self,
 			 unsigned long action, void *hcpu)
 {
@@ -4208,10 +4208,9 @@ void __init rcu_init(void)
 	 * this is called early in boot, before either interrupts
 	 * or the scheduler are operational.
 	 */
-	cpu_notifier(rcu_cpu_notify, 0);
 	pm_notifier(rcu_pm_notify, 0);
 	for_each_online_cpu(cpu)
-		rcu_cpu_notify(NULL, CPU_UP_PREPARE, (void *)(long)cpu);
+		rcutree_prepare_cpu(cpu);
 }
 
 #include "tree_exp.h"

[tip:smp/hotplug] rcu: Convert rcutree to hotplug state machine

[tip-bot for Thomas Gleixner]

Commit-ID:  4df8374254ea9294dfe4b8c447a1b7eddc543dbf
Gitweb:     http://git.kernel.org/tip/4df8374254ea9294dfe4b8c447a1b7eddc543dbf
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:03 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:44 +0200

rcu: Convert rcutree to hotplug state machine

Straight forward conversion to the state machine. Though the question arises
whether this needs really all these state transitions to work.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153337.982013161@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 include/linux/cpuhotplug.h |   3 ++
 include/linux/rcutiny.h    |   7 +++
 include/linux/rcutree.h    |   7 +++
 kernel/cpu.c               |  14 ++++++
 kernel/rcu/tree.c          | 105 ++++++++++++++++++++++-----------------------
 5 files changed, 83 insertions(+), 53 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 544b556..201a2e2 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,11 +20,13 @@ enum cpuhp_state {
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
 	CPUHP_TIMERS_DEAD,
+	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_BRINGUP_CPU,
 	CPUHP_AP_IDLE_DEAD,
 	CPUHP_AP_OFFLINE,
 	CPUHP_AP_SCHED_STARTING,
+	CPUHP_AP_RCUTREE_DYING,
 	CPUHP_AP_IRQ_GIC_STARTING,
 	CPUHP_AP_IRQ_GICV3_STARTING,
 	CPUHP_AP_IRQ_HIP04_STARTING,
@@ -80,6 +82,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,
 	CPUHP_AP_PERF_ARM_CCN_ONLINE,
 	CPUHP_AP_WORKQUEUE_ONLINE,
+	CPUHP_AP_RCUTREE_ONLINE,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
 	CPUHP_AP_ONLINE_DYN_END		= CPUHP_AP_ONLINE_DYN + 30,
diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h
index 93aea75..ac81e40 100644
--- a/include/linux/rcutiny.h
+++ b/include/linux/rcutiny.h
@@ -243,4 +243,11 @@ static inline void rcu_all_qs(void)
 	barrier(); /* Avoid RCU read-side critical sections leaking across. */
 }
 
+/* RCUtree hotplug events */
+#define rcutree_prepare_cpu      NULL
+#define rcutree_online_cpu       NULL
+#define rcutree_offline_cpu      NULL
+#define rcutree_dead_cpu         NULL
+#define rcutree_dying_cpu        NULL
+
 #endif /* __LINUX_RCUTINY_H */
diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h
index 5043cb8..63a4e4c 100644
--- a/include/linux/rcutree.h
+++ b/include/linux/rcutree.h
@@ -111,4 +111,11 @@ bool rcu_is_watching(void);
 
 void rcu_all_qs(void);
 
+/* RCUtree hotplug events */
+int rcutree_prepare_cpu(unsigned int cpu);
+int rcutree_online_cpu(unsigned int cpu);
+int rcutree_offline_cpu(unsigned int cpu);
+int rcutree_dead_cpu(unsigned int cpu);
+int rcutree_dying_cpu(unsigned int cpu);
+
 #endif /* __LINUX_RCUTREE_H */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 008e2fd..f24f459 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1205,6 +1205,11 @@ static struct cpuhp_step cpuhp_bp_states[] = {
 		.startup = NULL,
 		.teardown = timers_dead_cpu,
 	},
+	[CPUHP_RCUTREE_PREP] = {
+		.name = "RCU-tree prepare",
+		.startup = rcutree_prepare_cpu,
+		.teardown = rcutree_dead_cpu,
+	},
 	/*
 	 * Preparatory and dead notifiers. Will be replaced once the notifiers
 	 * are converted to states.
@@ -1263,6 +1268,10 @@ static struct cpuhp_step cpuhp_ap_states[] = {
 		.startup		= sched_cpu_starting,
 		.teardown		= sched_cpu_dying,
 	},
+	[CPUHP_AP_RCUTREE_DYING] = {
+		.startup = NULL,
+		.teardown = rcutree_dying_cpu,
+	},
 	/*
 	 * Low level startup/teardown notifiers. Run with interrupts
 	 * disabled. Will be removed once the notifiers are converted to
@@ -1296,6 +1305,11 @@ static struct cpuhp_step cpuhp_ap_states[] = {
 		.startup = workqueue_online_cpu,
 		.teardown = workqueue_offline_cpu,
 	},
+	[CPUHP_AP_RCUTREE_ONLINE] = {
+		.name = "RCU-tree online",
+		.startup = rcutree_online_cpu,
+		.teardown = rcutree_offline_cpu,
+	},
 
 	/*
 	 * Online/down_prepare notifiers. Will be removed once the notifiers
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index f433959..5d80925 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -1073,11 +1073,11 @@ EXPORT_SYMBOL_GPL(rcu_is_watching);
  * offline to continue to use RCU for one jiffy after marking itself
  * offline in the cpu_online_mask.  This leniency is necessary given the
  * non-atomic nature of the online and offline processing, for example,
- * the fact that a CPU enters the scheduler after completing the CPU_DYING
- * notifiers.
+ * the fact that a CPU enters the scheduler after completing the teardown
+ * of the CPU.
  *
- * This is also why RCU internally marks CPUs online during the
- * CPU_UP_PREPARE phase and offline during the CPU_DEAD phase.
+ * This is also why RCU internally marks CPUs online during in the
+ * preparation phase and offline after the CPU has been taken down.
  *
  * Disable checking if in an NMI handler because we cannot safely report
  * errors from NMI handlers anyway.
@@ -3806,12 +3806,58 @@ rcu_init_percpu_data(int cpu, struct rcu_state *rsp)
 	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 }
 
-static void rcu_prepare_cpu(int cpu)
+int rcutree_prepare_cpu(unsigned int cpu)
 {
 	struct rcu_state *rsp;
 
 	for_each_rcu_flavor(rsp)
 		rcu_init_percpu_data(cpu, rsp);
+
+	rcu_prepare_kthreads(cpu);
+	rcu_spawn_all_nocb_kthreads(cpu);
+
+	return 0;
+}
+
+static void rcutree_affinity_setting(unsigned int cpu, int outgoing)
+{
+	struct rcu_data *rdp = per_cpu_ptr(rcu_state_p->rda, cpu);
+
+	rcu_boost_kthread_setaffinity(rdp->mynode, outgoing);
+}
+
+int rcutree_online_cpu(unsigned int cpu)
+{
+	sync_sched_exp_online_cleanup(cpu);
+	rcutree_affinity_setting(cpu, -1);
+	return 0;
+}
+
+int rcutree_offline_cpu(unsigned int cpu)
+{
+	rcutree_affinity_setting(cpu, cpu);
+	return 0;
+}
+
+
+int rcutree_dying_cpu(unsigned int cpu)
+{
+	struct rcu_state *rsp;
+
+	for_each_rcu_flavor(rsp)
+		rcu_cleanup_dying_cpu(rsp);
+	return 0;
+}
+
+int rcutree_dead_cpu(unsigned int cpu)
+{
+	struct rcu_state *rsp;
+
+	for_each_rcu_flavor(rsp) {
+		rcu_cleanup_dead_cpu(cpu, rsp);
+		do_nocb_deferred_wakeup(per_cpu_ptr(rsp->rda, cpu));
+	}
+	return 0;
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -3851,52 +3897,6 @@ void rcu_report_dead(unsigned int cpu)
 }
 #endif
 
-/*
- * Handle CPU online/offline notification events.
- */
-int rcu_cpu_notify(struct notifier_block *self,
-		   unsigned long action, void *hcpu)
-{
-	long cpu = (long)hcpu;
-	struct rcu_data *rdp = per_cpu_ptr(rcu_state_p->rda, cpu);
-	struct rcu_node *rnp = rdp->mynode;
-	struct rcu_state *rsp;
-
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		rcu_prepare_cpu(cpu);
-		rcu_prepare_kthreads(cpu);
-		rcu_spawn_all_nocb_kthreads(cpu);
-		break;
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-		sync_sched_exp_online_cleanup(cpu);
-		rcu_boost_kthread_setaffinity(rnp, -1);
-		break;
-	case CPU_DOWN_PREPARE:
-		rcu_boost_kthread_setaffinity(rnp, cpu);
-		break;
-	case CPU_DYING:
-	case CPU_DYING_FROZEN:
-		for_each_rcu_flavor(rsp)
-			rcu_cleanup_dying_cpu(rsp);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
-		for_each_rcu_flavor(rsp) {
-			rcu_cleanup_dead_cpu(cpu, rsp);
-			do_nocb_deferred_wakeup(per_cpu_ptr(rsp->rda, cpu));
-		}
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_OK;
-}
-
 static int rcu_pm_notify(struct notifier_block *self,
 			 unsigned long action, void *hcpu)
 {
@@ -4208,10 +4208,9 @@ void __init rcu_init(void)
 	 * this is called early in boot, before either interrupts
 	 * or the scheduler are operational.
 	 */
-	cpu_notifier(rcu_cpu_notify, 0);
 	pm_notifier(rcu_pm_notify, 0);
 	for_each_online_cpu(cpu)
-		rcu_cpu_notify(NULL, CPU_UP_PREPARE, (void *)(long)cpu);
+		rcutree_prepare_cpu(cpu);
 }
 
 #include "tree_exp.h"

[patch V2 62/67] [PATCH] clocksource: arm_global_timer: Convert to hotplug state machine.

[Anna-Maria Gleixner]

[-- Attachment #1: 0032-clocksource-arm_global_timer-Convert-to-hotplug-stat.patch --]
[-- Type: text/plain, Size: 3249 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Srinivas Kandagatla <srinivas.kandagatla@gmail.com>
Cc: Maxime Coquelin <maxime.coquelin@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: kernel@stlinux.com
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/arm_global_timer.c |   39 ++++++++-------------------------
 include/linux/cpuhotplug.h             |    1 
 2 files changed, 11 insertions(+), 29 deletions(-)

--- a/drivers/clocksource/arm_global_timer.c
+++ b/drivers/clocksource/arm_global_timer.c
@@ -165,9 +165,9 @@ static irqreturn_t gt_clockevent_interru
 	return IRQ_HANDLED;
 }
 
-static int gt_clockevents_init(struct clock_event_device *clk)
+static int gt_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *clk = this_cpu_ptr(gt_evt);
 
 	clk->name = "arm_global_timer";
 	clk->features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT |
@@ -186,10 +186,13 @@ static int gt_clockevents_init(struct cl
 	return 0;
 }
 
-static void gt_clockevents_stop(struct clock_event_device *clk)
+static int gt_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *clk = this_cpu_ptr(gt_evt);
+
 	gt_clockevent_shutdown(clk);
 	disable_percpu_irq(clk->irq);
+	return 0;
 }
 
 static cycle_t gt_clocksource_read(struct clocksource *cs)
@@ -252,24 +255,6 @@ static int __init gt_clocksource_init(vo
 	return clocksource_register_hz(&gt_clocksource, gt_clk_rate);
 }
 
-static int gt_cpu_notify(struct notifier_block *self, unsigned long action,
-			 void *hcpu)
-{
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		gt_clockevents_init(this_cpu_ptr(gt_evt));
-		break;
-	case CPU_DYING:
-		gt_clockevents_stop(this_cpu_ptr(gt_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-static struct notifier_block gt_cpu_nb = {
-	.notifier_call = gt_cpu_notify,
-};
-
 static int __init global_timer_of_register(struct device_node *np)
 {
 	struct clk *gt_clk;
@@ -325,18 +310,14 @@ static int __init global_timer_of_regist
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&gt_cpu_nb);
-	if (err) {
-		pr_warn("global-timer: unable to register cpu notifier.\n");
-		goto out_irq;
-	}
-
-	/* Immediately configure the timer on the boot CPU */
+	/* Register and immediately configure the timer on the boot CPU */
 	err = gt_clocksource_init();
 	if (err)
 		goto out_irq;
 	
-	err = gt_clockevents_init(this_cpu_ptr(gt_evt));
+	err = cpuhp_setup_state(CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
+				"AP_ARM_GLOBAL_TIMER_STARTING",
+				gt_starting_cpu, gt_dying_cpu);
 	if (err)
 		goto out_irq;
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -48,6 +48,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+	CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,

[tip:smp/hotplug] clocksource/arm_global_timer: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  b8a12296ac812f829bd9392346b01b1872d8118b
Gitweb:     http://git.kernel.org/tip/b8a12296ac812f829bd9392346b01b1872d8118b
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:04 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:44 +0200

clocksource/arm_global_timer: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Maxime Coquelin <maxime.coquelin@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srinivas Kandagatla <srinivas.kandagatla@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kernel@stlinux.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153338.062741642@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/arm_global_timer.c | 39 +++++++++-------------------------
 include/linux/cpuhotplug.h             |  1 +
 2 files changed, 11 insertions(+), 29 deletions(-)

diff --git a/drivers/clocksource/arm_global_timer.c b/drivers/clocksource/arm_global_timer.c
index 2a9ceb6..8da0329 100644
--- a/drivers/clocksource/arm_global_timer.c
+++ b/drivers/clocksource/arm_global_timer.c
@@ -165,9 +165,9 @@ static irqreturn_t gt_clockevent_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static int gt_clockevents_init(struct clock_event_device *clk)
+static int gt_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *clk = this_cpu_ptr(gt_evt);
 
 	clk->name = "arm_global_timer";
 	clk->features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT |
@@ -186,10 +186,13 @@ static int gt_clockevents_init(struct clock_event_device *clk)
 	return 0;
 }
 
-static void gt_clockevents_stop(struct clock_event_device *clk)
+static int gt_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *clk = this_cpu_ptr(gt_evt);
+
 	gt_clockevent_shutdown(clk);
 	disable_percpu_irq(clk->irq);
+	return 0;
 }
 
 static cycle_t gt_clocksource_read(struct clocksource *cs)
@@ -252,24 +255,6 @@ static int __init gt_clocksource_init(void)
 	return clocksource_register_hz(&gt_clocksource, gt_clk_rate);
 }
 
-static int gt_cpu_notify(struct notifier_block *self, unsigned long action,
-			 void *hcpu)
-{
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		gt_clockevents_init(this_cpu_ptr(gt_evt));
-		break;
-	case CPU_DYING:
-		gt_clockevents_stop(this_cpu_ptr(gt_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-static struct notifier_block gt_cpu_nb = {
-	.notifier_call = gt_cpu_notify,
-};
-
 static int __init global_timer_of_register(struct device_node *np)
 {
 	struct clk *gt_clk;
@@ -325,18 +310,14 @@ static int __init global_timer_of_register(struct device_node *np)
 		goto out_free;
 	}
 
-	err = register_cpu_notifier(&gt_cpu_nb);
-	if (err) {
-		pr_warn("global-timer: unable to register cpu notifier.\n");
-		goto out_irq;
-	}
-
-	/* Immediately configure the timer on the boot CPU */
+	/* Register and immediately configure the timer on the boot CPU */
 	err = gt_clocksource_init();
 	if (err)
 		goto out_irq;
 	
-	err = gt_clockevents_init(this_cpu_ptr(gt_evt));
+	err = cpuhp_setup_state(CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
+				"AP_ARM_GLOBAL_TIMER_STARTING",
+				gt_starting_cpu, gt_dying_cpu);
 	if (err)
 		goto out_irq;
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 201a2e2..d24b0fc 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -47,6 +47,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_STARTING,
 	CPUHP_AP_ARM_L2X0_STARTING,
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
+	CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,

[patch V2 63/67] clocksource/exynos_mct: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0034-clocksource-exynos_mct-Convert-to-hotplug-state-mach.patch --]
[-- Type: text/plain, Size: 3514 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Kukjin Kim <kgene@kernel.org>
Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: linux-samsung-soc@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/exynos_mct.c |   46 ++++++++++-----------------------------
 include/linux/cpuhotplug.h       |    1 
 2 files changed, 13 insertions(+), 34 deletions(-)

--- a/drivers/clocksource/exynos_mct.c
+++ b/drivers/clocksource/exynos_mct.c
@@ -443,10 +443,11 @@ static irqreturn_t exynos4_mct_tick_isr(
 	return IRQ_HANDLED;
 }
 
-static int exynos4_local_timer_setup(struct mct_clock_event_device *mevt)
+static int exynos4_mct_starting_cpu(unsigned int cpu)
 {
+	struct mct_clock_event_device *mevt =
+		per_cpu_ptr(&percpu_mct_tick, cpu);
 	struct clock_event_device *evt = &mevt->evt;
-	unsigned int cpu = smp_processor_id();
 
 	mevt->base = EXYNOS4_MCT_L_BASE(cpu);
 	snprintf(mevt->name, sizeof(mevt->name), "mct_tick%d", cpu);
@@ -480,8 +481,10 @@ static int exynos4_local_timer_setup(str
 	return 0;
 }
 
-static void exynos4_local_timer_stop(struct mct_clock_event_device *mevt)
+static int exynos4_mct_dying_cpu(unsigned int cpu)
 {
+	struct mct_clock_event_device *mevt =
+		per_cpu_ptr(&percpu_mct_tick, cpu);
 	struct clock_event_device *evt = &mevt->evt;
 
 	evt->set_state_shutdown(evt);
@@ -491,39 +494,12 @@ static void exynos4_local_timer_stop(str
 	} else {
 		disable_percpu_irq(mct_irqs[MCT_L0_IRQ]);
 	}
+	return 0;
 }
 
-static int exynos4_mct_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	struct mct_clock_event_device *mevt;
-
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		mevt = this_cpu_ptr(&percpu_mct_tick);
-		exynos4_local_timer_setup(mevt);
-		break;
-	case CPU_DYING:
-		mevt = this_cpu_ptr(&percpu_mct_tick);
-		exynos4_local_timer_stop(mevt);
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block exynos4_mct_cpu_nb = {
-	.notifier_call = exynos4_mct_cpu_notify,
-};
-
 static int __init exynos4_timer_resources(struct device_node *np, void __iomem *base)
 {
 	int err, cpu;
-	struct mct_clock_event_device *mevt = this_cpu_ptr(&percpu_mct_tick);
 	struct clk *mct_clk, *tick_clk;
 
 	tick_clk = np ? of_clk_get_by_name(np, "fin_pll") :
@@ -570,12 +546,14 @@ static int __init exynos4_timer_resource
 		}
 	}
 
-	err = register_cpu_notifier(&exynos4_mct_cpu_nb);
+	/* Install hotplug callbacks which configure the timer on this CPU */
+	err = cpuhp_setup_state(CPUHP_AP_EXYNOS4_MCT_TIMER_STARTING,
+				"AP_EXYNOS4_MCT_TIMER_STARTING",
+				exynos4_mct_starting_cpu,
+				exynos4_mct_dying_cpu);
 	if (err)
 		goto out_irq;
 
-	/* Immediately configure the timer on the boot CPU */
-	exynos4_local_timer_setup(mevt);
 	return 0;
 
 out_irq:
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -50,6 +50,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_EXYNOS4_MCT_TIMER_STARTING,
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,

[tip:smp/hotplug] clocksource/exynos_mct: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  d11b3a60fcc8c6e44dd370f1cf2738eb5bec7353
Gitweb:     http://git.kernel.org/tip/d11b3a60fcc8c6e44dd370f1cf2738eb5bec7353
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:05 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:44 +0200

clocksource/exynos_mct: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: Kukjin Kim <kgene@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-samsung-soc@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153338.147940411@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/exynos_mct.c | 46 +++++++++++-----------------------------
 include/linux/cpuhotplug.h       |  1 +
 2 files changed, 13 insertions(+), 34 deletions(-)

diff --git a/drivers/clocksource/exynos_mct.c b/drivers/clocksource/exynos_mct.c
index 0d18dd4b..41840d0 100644
--- a/drivers/clocksource/exynos_mct.c
+++ b/drivers/clocksource/exynos_mct.c
@@ -443,10 +443,11 @@ static irqreturn_t exynos4_mct_tick_isr(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static int exynos4_local_timer_setup(struct mct_clock_event_device *mevt)
+static int exynos4_mct_starting_cpu(unsigned int cpu)
 {
+	struct mct_clock_event_device *mevt =
+		per_cpu_ptr(&percpu_mct_tick, cpu);
 	struct clock_event_device *evt = &mevt->evt;
-	unsigned int cpu = smp_processor_id();
 
 	mevt->base = EXYNOS4_MCT_L_BASE(cpu);
 	snprintf(mevt->name, sizeof(mevt->name), "mct_tick%d", cpu);
@@ -480,8 +481,10 @@ static int exynos4_local_timer_setup(struct mct_clock_event_device *mevt)
 	return 0;
 }
 
-static void exynos4_local_timer_stop(struct mct_clock_event_device *mevt)
+static int exynos4_mct_dying_cpu(unsigned int cpu)
 {
+	struct mct_clock_event_device *mevt =
+		per_cpu_ptr(&percpu_mct_tick, cpu);
 	struct clock_event_device *evt = &mevt->evt;
 
 	evt->set_state_shutdown(evt);
@@ -491,39 +494,12 @@ static void exynos4_local_timer_stop(struct mct_clock_event_device *mevt)
 	} else {
 		disable_percpu_irq(mct_irqs[MCT_L0_IRQ]);
 	}
+	return 0;
 }
 
-static int exynos4_mct_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	struct mct_clock_event_device *mevt;
-
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		mevt = this_cpu_ptr(&percpu_mct_tick);
-		exynos4_local_timer_setup(mevt);
-		break;
-	case CPU_DYING:
-		mevt = this_cpu_ptr(&percpu_mct_tick);
-		exynos4_local_timer_stop(mevt);
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block exynos4_mct_cpu_nb = {
-	.notifier_call = exynos4_mct_cpu_notify,
-};
-
 static int __init exynos4_timer_resources(struct device_node *np, void __iomem *base)
 {
 	int err, cpu;
-	struct mct_clock_event_device *mevt = this_cpu_ptr(&percpu_mct_tick);
 	struct clk *mct_clk, *tick_clk;
 
 	tick_clk = np ? of_clk_get_by_name(np, "fin_pll") :
@@ -570,12 +546,14 @@ static int __init exynos4_timer_resources(struct device_node *np, void __iomem *
 		}
 	}
 
-	err = register_cpu_notifier(&exynos4_mct_cpu_nb);
+	/* Install hotplug callbacks which configure the timer on this CPU */
+	err = cpuhp_setup_state(CPUHP_AP_EXYNOS4_MCT_TIMER_STARTING,
+				"AP_EXYNOS4_MCT_TIMER_STARTING",
+				exynos4_mct_starting_cpu,
+				exynos4_mct_dying_cpu);
 	if (err)
 		goto out_irq;
 
-	/* Immediately configure the timer on the boot CPU */
-	exynos4_local_timer_setup(mevt);
 	return 0;
 
 out_irq:
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index d24b0fc..7ad8a7e7 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -49,6 +49,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_ARCH_TIMER_STARTING,
 	CPUHP_AP_ARM_GLOBAL_TIMER_STARTING,
 	CPUHP_AP_DUMMY_TIMER_STARTING,
+	CPUHP_AP_EXYNOS4_MCT_TIMER_STARTING,
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,

[patch V2 64/67] clocksource/armada-370-xp: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0037-clocksource-armada-370-xp-Convert-to-hotplug-state-m.patch --]
[-- Type: text/plain, Size: 3105 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.


Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/time-armada-370-xp.c |   41 ++++++++-----------------------
 include/linux/cpuhotplug.h               |    1 
 2 files changed, 12 insertions(+), 30 deletions(-)

--- a/drivers/clocksource/time-armada-370-xp.c
+++ b/drivers/clocksource/time-armada-370-xp.c
@@ -170,10 +170,10 @@ static irqreturn_t armada_370_xp_timer_i
 /*
  * Setup the local clock events for a CPU.
  */
-static int armada_370_xp_timer_setup(struct clock_event_device *evt)
+static int armada_370_xp_timer_starting_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(armada_370_xp_evt, cpu);
 	u32 clr = 0, set = 0;
-	int cpu = smp_processor_id();
 
 	if (timer25Mhz)
 		set = TIMER0_25MHZ;
@@ -200,35 +200,15 @@ static int armada_370_xp_timer_setup(str
 	return 0;
 }
 
-static void armada_370_xp_timer_stop(struct clock_event_device *evt)
+static int armada_370_xp_timer_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(armada_370_xp_evt, cpu);
+
 	evt->set_state_shutdown(evt);
 	disable_percpu_irq(evt->irq);
+	return 0;
 }
 
-static int armada_370_xp_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		armada_370_xp_timer_setup(this_cpu_ptr(armada_370_xp_evt));
-		break;
-	case CPU_DYING:
-		armada_370_xp_timer_stop(this_cpu_ptr(armada_370_xp_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block armada_370_xp_timer_cpu_nb = {
-	.notifier_call = armada_370_xp_timer_cpu_notify,
-};
-
 static u32 timer0_ctrl_reg, timer0_local_ctrl_reg;
 
 static int armada_370_xp_timer_suspend(void)
@@ -322,8 +302,6 @@ static int __init armada_370_xp_timer_co
 		return res;
 	}
 
-	register_cpu_notifier(&armada_370_xp_timer_cpu_nb);
-
 	armada_370_xp_evt = alloc_percpu(struct clock_event_device);
 	if (!armada_370_xp_evt)
 		return -ENOMEM;
@@ -341,9 +319,12 @@ static int __init armada_370_xp_timer_co
 		return res;
 	}
 
-	res = armada_370_xp_timer_setup(this_cpu_ptr(armada_370_xp_evt));
+	res = cpuhp_setup_state(CPUHP_AP_ARMADA_TIMER_STARTING,
+				"AP_ARMADA_TIMER_STARTING",
+				armada_370_xp_timer_starting_cpu,
+				armada_370_xp_timer_dying_cpu);
 	if (res) {
-		pr_err("Failed to setup timer");
+		pr_err("Failed to setup hotplug state and timer");
 		return res;
 	}
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -54,6 +54,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
+	CPUHP_AP_ARMADA_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,

[tip:smp/hotplug] clocksource/armada-370-xp: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  2c48fef74ca42d05c3efd4345b8a068ebf479a5f
Gitweb:     http://git.kernel.org/tip/2c48fef74ca42d05c3efd4345b8a068ebf479a5f
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:06 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:45 +0200

clocksource/armada-370-xp: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153338.229913786@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/time-armada-370-xp.c | 41 +++++++++-----------------------
 include/linux/cpuhotplug.h               |  1 +
 2 files changed, 12 insertions(+), 30 deletions(-)

diff --git a/drivers/clocksource/time-armada-370-xp.c b/drivers/clocksource/time-armada-370-xp.c
index 20ec066..719b478 100644
--- a/drivers/clocksource/time-armada-370-xp.c
+++ b/drivers/clocksource/time-armada-370-xp.c
@@ -170,10 +170,10 @@ static irqreturn_t armada_370_xp_timer_interrupt(int irq, void *dev_id)
 /*
  * Setup the local clock events for a CPU.
  */
-static int armada_370_xp_timer_setup(struct clock_event_device *evt)
+static int armada_370_xp_timer_starting_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(armada_370_xp_evt, cpu);
 	u32 clr = 0, set = 0;
-	int cpu = smp_processor_id();
 
 	if (timer25Mhz)
 		set = TIMER0_25MHZ;
@@ -200,35 +200,15 @@ static int armada_370_xp_timer_setup(struct clock_event_device *evt)
 	return 0;
 }
 
-static void armada_370_xp_timer_stop(struct clock_event_device *evt)
+static int armada_370_xp_timer_dying_cpu(unsigned int cpu)
 {
+	struct clock_event_device *evt = per_cpu_ptr(armada_370_xp_evt, cpu);
+
 	evt->set_state_shutdown(evt);
 	disable_percpu_irq(evt->irq);
+	return 0;
 }
 
-static int armada_370_xp_timer_cpu_notify(struct notifier_block *self,
-					   unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		armada_370_xp_timer_setup(this_cpu_ptr(armada_370_xp_evt));
-		break;
-	case CPU_DYING:
-		armada_370_xp_timer_stop(this_cpu_ptr(armada_370_xp_evt));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block armada_370_xp_timer_cpu_nb = {
-	.notifier_call = armada_370_xp_timer_cpu_notify,
-};
-
 static u32 timer0_ctrl_reg, timer0_local_ctrl_reg;
 
 static int armada_370_xp_timer_suspend(void)
@@ -322,8 +302,6 @@ static int __init armada_370_xp_timer_common_init(struct device_node *np)
 		return res;
 	}
 
-	register_cpu_notifier(&armada_370_xp_timer_cpu_nb);
-
 	armada_370_xp_evt = alloc_percpu(struct clock_event_device);
 	if (!armada_370_xp_evt)
 		return -ENOMEM;
@@ -341,9 +319,12 @@ static int __init armada_370_xp_timer_common_init(struct device_node *np)
 		return res;
 	}
 
-	res = armada_370_xp_timer_setup(this_cpu_ptr(armada_370_xp_evt));
+	res = cpuhp_setup_state(CPUHP_AP_ARMADA_TIMER_STARTING,
+				"AP_ARMADA_TIMER_STARTING",
+				armada_370_xp_timer_starting_cpu,
+				armada_370_xp_timer_dying_cpu);
 	if (res) {
-		pr_err("Failed to setup timer");
+		pr_err("Failed to setup hotplug state and timer");
 		return res;
 	}
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 7ad8a7e7..ee7f750 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -53,6 +53,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARM_TWD_STARTING,
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
+	CPUHP_AP_ARMADA_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,

[patch V2 65/67] clocksource/atlas7: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: 0038-clocksource-atlas7-Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2993 bytes --]

From: Richard Cochran <rcochran@linutronix.de>

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Barry Song <baohua@kernel.org>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 drivers/clocksource/timer-atlas7.c |   41 ++++++++-----------------------------
 include/linux/cpuhotplug.h         |    1 
 2 files changed, 10 insertions(+), 32 deletions(-)

--- a/drivers/clocksource/timer-atlas7.c
+++ b/drivers/clocksource/timer-atlas7.c
@@ -172,9 +172,9 @@ static struct irqaction sirfsoc_timer1_i
 	.handler = sirfsoc_timer_interrupt,
 };
 
-static int sirfsoc_local_timer_setup(struct clock_event_device *ce)
+static int sirfsoc_local_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *ce = per_cpu_ptr(sirfsoc_clockevent, cpu);
 	struct irqaction *action;
 
 	if (cpu == 0)
@@ -203,50 +203,27 @@ static int sirfsoc_local_timer_setup(str
 	return 0;
 }
 
-static void sirfsoc_local_timer_stop(struct clock_event_device *ce)
+static int sirfsoc_local_timer_dying_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
-
 	sirfsoc_timer_count_disable(1);
 
 	if (cpu == 0)
 		remove_irq(sirfsoc_timer_irq.irq, &sirfsoc_timer_irq);
 	else
 		remove_irq(sirfsoc_timer1_irq.irq, &sirfsoc_timer1_irq);
+	return 0;
 }
 
-static int sirfsoc_cpu_notify(struct notifier_block *self,
-			      unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		sirfsoc_local_timer_setup(this_cpu_ptr(sirfsoc_clockevent));
-		break;
-	case CPU_DYING:
-		sirfsoc_local_timer_stop(this_cpu_ptr(sirfsoc_clockevent));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block sirfsoc_cpu_nb = {
-	.notifier_call = sirfsoc_cpu_notify,
-};
-
 static int __init sirfsoc_clockevent_init(void)
 {
 	sirfsoc_clockevent = alloc_percpu(struct clock_event_device);
 	BUG_ON(!sirfsoc_clockevent);
 
-	BUG_ON(register_cpu_notifier(&sirfsoc_cpu_nb));
-
-	/* Immediately configure the timer on the boot CPU */
-	return sirfsoc_local_timer_setup(this_cpu_ptr(sirfsoc_clockevent));
+	/* Install and invoke hotplug callbacks */
+	return cpuhp_setup_state(CPUHP_AP_MARCO_TIMER_STARTING,
+				 "AP_MARCO_TIMER_STARTING",
+				 sirfsoc_local_timer_starting_cpu,
+				 sirfsoc_local_timer_dying_cpu);
 }
 
 /* initialize the kernel jiffy timer source */
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -55,6 +55,7 @@ enum cpuhp_state {
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_ARMADA_TIMER_STARTING,
+	CPUHP_AP_MARCO_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,

[tip:smp/hotplug] clocksource/atlas7: Convert to hotplug state machine

[tip-bot for Richard Cochran]

Commit-ID:  eb0a9d8c672dc01db41352afa646405d035ee7a4
Gitweb:     http://git.kernel.org/tip/eb0a9d8c672dc01db41352afa646405d035ee7a4
Author:     Richard Cochran <rcochran@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:07 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:45 +0200

clocksource/atlas7: Convert to hotplug state machine

Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Barry Song <baohua@kernel.org>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153338.310333816@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 drivers/clocksource/timer-atlas7.c | 41 +++++++++-----------------------------
 include/linux/cpuhotplug.h         |  1 +
 2 files changed, 10 insertions(+), 32 deletions(-)

diff --git a/drivers/clocksource/timer-atlas7.c b/drivers/clocksource/timer-atlas7.c
index 90f8fbc..4334e03 100644
--- a/drivers/clocksource/timer-atlas7.c
+++ b/drivers/clocksource/timer-atlas7.c
@@ -172,9 +172,9 @@ static struct irqaction sirfsoc_timer1_irq = {
 	.handler = sirfsoc_timer_interrupt,
 };
 
-static int sirfsoc_local_timer_setup(struct clock_event_device *ce)
+static int sirfsoc_local_timer_starting_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
+	struct clock_event_device *ce = per_cpu_ptr(sirfsoc_clockevent, cpu);
 	struct irqaction *action;
 
 	if (cpu == 0)
@@ -203,50 +203,27 @@ static int sirfsoc_local_timer_setup(struct clock_event_device *ce)
 	return 0;
 }
 
-static void sirfsoc_local_timer_stop(struct clock_event_device *ce)
+static int sirfsoc_local_timer_dying_cpu(unsigned int cpu)
 {
-	int cpu = smp_processor_id();
-
 	sirfsoc_timer_count_disable(1);
 
 	if (cpu == 0)
 		remove_irq(sirfsoc_timer_irq.irq, &sirfsoc_timer_irq);
 	else
 		remove_irq(sirfsoc_timer1_irq.irq, &sirfsoc_timer1_irq);
+	return 0;
 }
 
-static int sirfsoc_cpu_notify(struct notifier_block *self,
-			      unsigned long action, void *hcpu)
-{
-	/*
-	 * Grab cpu pointer in each case to avoid spurious
-	 * preemptible warnings
-	 */
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		sirfsoc_local_timer_setup(this_cpu_ptr(sirfsoc_clockevent));
-		break;
-	case CPU_DYING:
-		sirfsoc_local_timer_stop(this_cpu_ptr(sirfsoc_clockevent));
-		break;
-	}
-
-	return NOTIFY_OK;
-}
-
-static struct notifier_block sirfsoc_cpu_nb = {
-	.notifier_call = sirfsoc_cpu_notify,
-};
-
 static int __init sirfsoc_clockevent_init(void)
 {
 	sirfsoc_clockevent = alloc_percpu(struct clock_event_device);
 	BUG_ON(!sirfsoc_clockevent);
 
-	BUG_ON(register_cpu_notifier(&sirfsoc_cpu_nb));
-
-	/* Immediately configure the timer on the boot CPU */
-	return sirfsoc_local_timer_setup(this_cpu_ptr(sirfsoc_clockevent));
+	/* Install and invoke hotplug callbacks */
+	return cpuhp_setup_state(CPUHP_AP_MARCO_TIMER_STARTING,
+				 "AP_MARCO_TIMER_STARTING",
+				 sirfsoc_local_timer_starting_cpu,
+				 sirfsoc_local_timer_dying_cpu);
 }
 
 /* initialize the kernel jiffy timer source */
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index ee7f750..5ac59f7 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -54,6 +54,7 @@ enum cpuhp_state {
 	CPUHP_AP_METAG_TIMER_STARTING,
 	CPUHP_AP_QCOM_TIMER_STARTING,
 	CPUHP_AP_ARMADA_TIMER_STARTING,
+	CPUHP_AP_MARCO_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,

[patch V2 66/67] ARC/time: Convert to hotplug state machine

[Anna-Maria Gleixner]

[-- Attachment #1: ARC-time--Convert-to-hotplug-state-machine.patch --]
[-- Type: text/plain, Size: 2996 bytes --]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

Install the callbacks via the state machine.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: linux-snps-arc@lists.infradead.org
---
 arch/arc/kernel/time.c     |   48 ++++++++++++++++-----------------------------
 include/linux/cpuhotplug.h |    1 
 2 files changed, 19 insertions(+), 30 deletions(-)

--- a/arch/arc/kernel/time.c
+++ b/arch/arc/kernel/time.c
@@ -296,30 +296,23 @@ static irqreturn_t timer_irq_handler(int
 	return IRQ_HANDLED;
 }
 
-static int arc_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+
+static int arc_timer_starting_cpu(unsigned int cpu)
 {
 	struct clock_event_device *evt = this_cpu_ptr(&arc_clockevent_device);
 
 	evt->cpumask = cpumask_of(smp_processor_id());
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		clockevents_config_and_register(evt, arc_timer_freq,
-						0, ULONG_MAX);
-		enable_percpu_irq(arc_timer_irq, 0);
-		break;
-	case CPU_DYING:
-		disable_percpu_irq(arc_timer_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	clockevents_config_and_register(evt, arc_timer_freq, 0, ARC_TIMER_MAX);
+	enable_percpu_irq(arc_timer_irq, 0);
+	return 0;
 }
 
-static struct notifier_block arc_timer_cpu_nb = {
-	.notifier_call = arc_timer_cpu_notify,
-};
+static int arc_timer_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(arc_timer_irq);
+	return 0;
+}
 
 /*
  * clockevent setup for boot CPU
@@ -329,12 +322,6 @@ static int __init arc_clockevent_setup(s
 	struct clock_event_device *evt = this_cpu_ptr(&arc_clockevent_device);
 	int ret;
 
-	ret = register_cpu_notifier(&arc_timer_cpu_nb);
-	if (ret) {
-		pr_err("Failed to register cpu notifier");
-		return ret;
-	}
-
 	arc_timer_irq = irq_of_parse_and_map(node, 0);
 	if (arc_timer_irq <= 0) {
 		pr_err("clockevent: missing irq");
@@ -347,11 +334,6 @@ static int __init arc_clockevent_setup(s
 		return ret;
 	}
 
-	evt->irq = arc_timer_irq;
-	evt->cpumask = cpumask_of(smp_processor_id());
-	clockevents_config_and_register(evt, arc_timer_freq,
-					0, ARC_TIMER_MAX);
-
 	/* Needs apriori irq_set_percpu_devid() done in intc map function */
 	ret = request_percpu_irq(arc_timer_irq, timer_irq_handler,
 				 "Timer0 (per-cpu-tick)", evt);
@@ -360,8 +342,14 @@ static int __init arc_clockevent_setup(s
 		return ret;
 	}
 
-	enable_percpu_irq(arc_timer_irq, 0);
-
+	ret = cpuhp_setup_state(CPUHP_AP_ARC_TIMER_STARTING,
+				"AP_ARC_TIMER_STARTING",
+				arc_timer_starting_cpu,
+				arc_timer_dying_cpu);
+	if (ret) {
+		pr_err("Failed to setup hotplug state");
+		return ret;
+	}
 	return 0;
 }
 
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -57,6 +57,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARMADA_TIMER_STARTING,
 	CPUHP_AP_MARCO_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
+	CPUHP_AP_ARC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,

[tip:smp/hotplug] ARC/time: Convert to hotplug state machine

[tip-bot for Anna-Maria Gleixner]

Commit-ID:  ecd8081f6fb9032c41e5c38885f06836da6ab455
Gitweb:     http://git.kernel.org/tip/ecd8081f6fb9032c41e5c38885f06836da6ab455
Author:     Anna-Maria Gleixner <anna-maria@linutronix.de>
AuthorDate: Wed, 13 Jul 2016 17:17:07 +0000
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 15 Jul 2016 10:41:46 +0200

ARC/time: Convert to hotplug state machine

Install the callbacks via the state machine.

Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: linux-snps-arc@lists.infradead.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160713153338.391826254@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/arc/kernel/time.c     | 48 +++++++++++++++++-----------------------------
 include/linux/cpuhotplug.h |  1 +
 2 files changed, 19 insertions(+), 30 deletions(-)

diff --git a/arch/arc/kernel/time.c b/arch/arc/kernel/time.c
index 98f22d2..f927b8d 100644
--- a/arch/arc/kernel/time.c
+++ b/arch/arc/kernel/time.c
@@ -296,30 +296,23 @@ static irqreturn_t timer_irq_handler(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static int arc_timer_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
+
+static int arc_timer_starting_cpu(unsigned int cpu)
 {
 	struct clock_event_device *evt = this_cpu_ptr(&arc_clockevent_device);
 
 	evt->cpumask = cpumask_of(smp_processor_id());
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_STARTING:
-		clockevents_config_and_register(evt, arc_timer_freq,
-						0, ULONG_MAX);
-		enable_percpu_irq(arc_timer_irq, 0);
-		break;
-	case CPU_DYING:
-		disable_percpu_irq(arc_timer_irq);
-		break;
-	}
-
-	return NOTIFY_OK;
+	clockevents_config_and_register(evt, arc_timer_freq, 0, ARC_TIMER_MAX);
+	enable_percpu_irq(arc_timer_irq, 0);
+	return 0;
 }
 
-static struct notifier_block arc_timer_cpu_nb = {
-	.notifier_call = arc_timer_cpu_notify,
-};
+static int arc_timer_dying_cpu(unsigned int cpu)
+{
+	disable_percpu_irq(arc_timer_irq);
+	return 0;
+}
 
 /*
  * clockevent setup for boot CPU
@@ -329,12 +322,6 @@ static int __init arc_clockevent_setup(struct device_node *node)
 	struct clock_event_device *evt = this_cpu_ptr(&arc_clockevent_device);
 	int ret;
 
-	ret = register_cpu_notifier(&arc_timer_cpu_nb);
-	if (ret) {
-		pr_err("Failed to register cpu notifier");
-		return ret;
-	}
-
 	arc_timer_irq = irq_of_parse_and_map(node, 0);
 	if (arc_timer_irq <= 0) {
 		pr_err("clockevent: missing irq");
@@ -347,11 +334,6 @@ static int __init arc_clockevent_setup(struct device_node *node)
 		return ret;
 	}
 
-	evt->irq = arc_timer_irq;
-	evt->cpumask = cpumask_of(smp_processor_id());
-	clockevents_config_and_register(evt, arc_timer_freq,
-					0, ARC_TIMER_MAX);
-
 	/* Needs apriori irq_set_percpu_devid() done in intc map function */
 	ret = request_percpu_irq(arc_timer_irq, timer_irq_handler,
 				 "Timer0 (per-cpu-tick)", evt);
@@ -360,8 +342,14 @@ static int __init arc_clockevent_setup(struct device_node *node)
 		return ret;
 	}
 
-	enable_percpu_irq(arc_timer_irq, 0);
-
+	ret = cpuhp_setup_state(CPUHP_AP_ARC_TIMER_STARTING,
+				"AP_ARC_TIMER_STARTING",
+				arc_timer_starting_cpu,
+				arc_timer_dying_cpu);
+	if (ret) {
+		pr_err("Failed to setup hotplug state");
+		return ret;
+	}
 	return 0;
 }
 
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 5ac59f7..09ef54b 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -56,6 +56,7 @@ enum cpuhp_state {
 	CPUHP_AP_ARMADA_TIMER_STARTING,
 	CPUHP_AP_MARCO_TIMER_STARTING,
 	CPUHP_AP_MIPS_GIC_TIMER_STARTING,
+	CPUHP_AP_ARC_TIMER_STARTING,
 	CPUHP_AP_KVM_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_INIT_STARTING,
 	CPUHP_AP_KVM_ARM_VGIC_STARTING,

[patch V2 67/67] cpuhotplug: Remove CPU_STARTING and CPU_DYING notifier

[Anna-Maria Gleixner]

[-- Attachment #1: 0059-cpuhotplug-Remove-CPU_STARTING-notifier.patch --]
[-- Type: text/plain, Size: 6156 bytes --]

From: Thomas Gleixner <tglx@linutronix.de>

All users are converted to the new state machine, remove CPU_STARTING and the
corresponding CPU_DYING.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
---
 arch/sparc/kernel/smp_32.c           |    2 --
 include/linux/cpu.h                  |   12 ------------
 include/linux/cpuhotplug.h           |    1 -
 kernel/cpu.c                         |   30 ++----------------------------
 tools/testing/radix-tree/linux/cpu.h |   13 -------------
 5 files changed, 2 insertions(+), 56 deletions(-)

--- a/arch/sparc/kernel/smp_32.c
+++ b/arch/sparc/kernel/smp_32.c
@@ -352,9 +352,7 @@ static void sparc_start_secondary(void *
 	preempt_disable();
 	cpu = smp_processor_id();
 
-	/* Invoke the CPU_STARTING notifier callbacks */
 	notify_cpu_starting(cpu);
-
 	arch_cpu_pre_online(arg);
 
 	/* Set the CPU in the cpu_online_mask */
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -61,17 +61,8 @@ struct notifier_block;
 #define CPU_DOWN_PREPARE	0x0005 /* CPU (unsigned)v going down */
 #define CPU_DOWN_FAILED		0x0006 /* CPU (unsigned)v NOT going down */
 #define CPU_DEAD		0x0007 /* CPU (unsigned)v dead */
-#define CPU_DYING		0x0008 /* CPU (unsigned)v not running any task,
-					* not handling interrupts, soon dead.
-					* Called on the dying cpu, interrupts
-					* are already disabled. Must not
-					* sleep, must not fail */
 #define CPU_POST_DEAD		0x0009 /* CPU (unsigned)v dead, cpu_hotplug
 					* lock is dropped */
-#define CPU_STARTING		0x000A /* CPU (unsigned)v soon running.
-					* Called on the new cpu, just before
-					* enabling interrupts. Must not sleep,
-					* must not fail */
 #define CPU_BROKEN		0x000B /* CPU (unsigned)v did not die properly,
 					* perhaps due to preemption. */
 
@@ -86,9 +77,6 @@ struct notifier_block;
 #define CPU_DOWN_PREPARE_FROZEN	(CPU_DOWN_PREPARE | CPU_TASKS_FROZEN)
 #define CPU_DOWN_FAILED_FROZEN	(CPU_DOWN_FAILED | CPU_TASKS_FROZEN)
 #define CPU_DEAD_FROZEN		(CPU_DEAD | CPU_TASKS_FROZEN)
-#define CPU_DYING_FROZEN	(CPU_DYING | CPU_TASKS_FROZEN)
-#define CPU_STARTING_FROZEN	(CPU_STARTING | CPU_TASKS_FROZEN)
-
 
 #ifdef CONFIG_SMP
 extern bool cpuhp_tasks_frozen;
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -69,7 +69,6 @@ enum cpuhp_state {
 	CPUHP_AP_LEDTRIG_STARTING,
 	CPUHP_AP_SMPCFD_DYING,
 	CPUHP_AP_X86_TBOOT_DYING,
-	CPUHP_AP_NOTIFY_STARTING,
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
 	CPUHP_AP_ONLINE_IDLE,
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -330,12 +330,6 @@ static int notify_online(unsigned int cp
 	return 0;
 }
 
-static int notify_starting(unsigned int cpu)
-{
-	cpu_notify(CPU_STARTING, cpu);
-	return 0;
-}
-
 static int bringup_wait_for_ap(unsigned int cpu)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
@@ -674,12 +668,6 @@ static int notify_down_prepare(unsigned
 	return err;
 }
 
-static int notify_dying(unsigned int cpu)
-{
-	cpu_notify(CPU_DYING, cpu);
-	return 0;
-}
-
 /* Take this CPU down. */
 static int take_cpu_down(void *_param)
 {
@@ -734,7 +722,7 @@ static int takedown_cpu(unsigned int cpu
 	BUG_ON(cpu_online(cpu));
 
 	/*
-	 * The migration_call() CPU_DYING callback will have removed all
+	 * The CPUHP_AP_SCHED_MIGRATE_DYING callback will have removed all
 	 * runnable tasks from the cpu, there's only the idle task left now
 	 * that the migration thread is done doing the stop_machine thing.
 	 *
@@ -787,7 +775,6 @@ void cpuhp_report_idle_dead(void)
 #define notify_down_prepare	NULL
 #define takedown_cpu		NULL
 #define notify_dead		NULL
-#define notify_dying		NULL
 #endif
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -877,10 +864,9 @@ EXPORT_SYMBOL(cpu_down);
 #endif /*CONFIG_HOTPLUG_CPU*/
 
 /**
- * notify_cpu_starting(cpu) - call the CPU_STARTING notifiers
+ * notify_cpu_starting(cpu) - Invoke the callbacks on the starting CPU
  * @cpu: cpu that just started
  *
- * This function calls the cpu_chain notifiers with CPU_STARTING.
  * It must be called by the arch code on the new cpu, before the new cpu
  * enables interrupts and before the "boot" cpu returns from __cpu_up().
  */
@@ -1270,18 +1256,6 @@ static struct cpuhp_step cpuhp_ap_states
 		.startup = NULL,
 		.teardown = rcutree_dying_cpu,
 	},
-	/*
-	 * Low level startup/teardown notifiers. Run with interrupts
-	 * disabled. Will be removed once the notifiers are converted to
-	 * states.
-	 */
-	[CPUHP_AP_NOTIFY_STARTING] = {
-		.name			= "notify:starting",
-		.startup		= notify_starting,
-		.teardown		= notify_dying,
-		.skip_onerr		= true,
-		.cant_stop		= true,
-	},
 	/* Entry state on starting. Interrupts enabled from here on. Transient
 	 * state for synchronsization */
 	[CPUHP_AP_ONLINE] = {
--- a/tools/testing/radix-tree/linux/cpu.h
+++ b/tools/testing/radix-tree/linux/cpu.h
@@ -7,19 +7,8 @@
 #define CPU_DOWN_PREPARE	0x0005 /* CPU (unsigned)v going down */
 #define CPU_DOWN_FAILED		0x0006 /* CPU (unsigned)v NOT going down */
 #define CPU_DEAD		0x0007 /* CPU (unsigned)v dead */
-#define CPU_DYING		0x0008 /* CPU (unsigned)v not running any task,
-					* not handling interrupts, soon dead.
-					* Called on the dying cpu, interrupts
-					* are already disabled. Must not
-					* sleep, must not fail */
 #define CPU_POST_DEAD		0x0009 /* CPU (unsigned)v dead, cpu_hotplug
 					* lock is dropped */
-#define CPU_STARTING		0x000A /* CPU (unsigned)v soon running.
-					* Called on the new cpu, just before
-					* enabling interrupts. Must not sleep,
-					* must not fail */
-#define CPU_DYING_IDLE		0x000B /* CPU (unsigned)v dying, reached
-					* idle loop. */
 #define CPU_BROKEN		0x000C /* CPU (unsigned)v did not die properly,
 					* perhaps due to preemption. */
 #define CPU_TASKS_FROZEN	0x0010
@@ -30,5 +19,3 @@
 #define CPU_DOWN_PREPARE_FROZEN (CPU_DOWN_PREPARE | CPU_TASKS_FROZEN)
 #define CPU_DOWN_FAILED_FROZEN  (CPU_DOWN_FAILED | CPU_TASKS_FROZEN)
 #define CPU_DEAD_FROZEN		(CPU_DEAD | CPU_TASKS_FROZEN)
-#define CPU_DYING_FROZEN	(CPU_DYING | CPU_TASKS_FROZEN)
-#define CPU_STARTING_FROZEN	(CPU_STARTING | CPU_TASKS_FROZEN)

Re: [patch V2 00/67] cpuhotplug: Convert all priority notifiers to the state machine

[Jason Cooper]

Anna-Maria,

On Wed, Jul 13, 2016 at 05:16:02PM -0000, Anna-Maria Gleixner wrote:

For (patches 3 - 7):

>  drivers/irqchip/irq-armada-370-xp.c           |   44 ++----
>  drivers/irqchip/irq-bcm2836.c                 |   34 +---
>  drivers/irqchip/irq-gic-v3.c                  |   22 +--
>  drivers/irqchip/irq-gic.c                     |   23 ---
>  drivers/irqchip/irq-hip04.c                   |   25 ---

Acked-by: Jason Cooper <jason@lakedaemon.net>

thx,

Jason.