cpu hotplug: add multi instance support

cpu hotplug: add multi instance support

[Sebastian Andrzej Siewior]

This series adds the multi instance support for a single hotplug state. We have
currently three drivers in tree which have their own list in order to handle
their individual instances.
There will be probably more of these drivers than less so we try to shift the
bits and pieces into the core code. The three ARM drivers are converted as part
of this series but only compile tested. The virtio-net was used to test patch
one and two.

Sebastian

[PATCH 1/6] cpu/hotplug: Rework callback invocation logic

[Sebastian Andrzej Siewior]

From: Thomas Gleixner <tglx@linutronix.de>

This is preparation for the following patch.
This rework here changes the arguments of cpuhp_invoke_callback(). It
passes now `state' and whether `startup' or `teardown' callback should
be invoked. The callback then is looked up by the function.

The following is a clanup of callers:
- cpuhp_issue_call() has one argument less
- struct cpuhp_cpu_state (which is used by the hotplug thread) gets also
  its callback removed. The decision if it is a single callback
  invocation moved to the `single' variable. Also a `bringup' variable
  has been added to distinguish between startup and teardown callback.
- take_cpu_down() needs to start one step earlier. We always get here
  via CPUHP_TEARDOWN_CPU callback. Before that change cpuhp_ap_states +
  CPUHP_TEARDOWN_CPU pointed to an empty entry because TEARDOWN is saved
  in bp_states for this reason. Now that we use cpuhp_get_step() to
  lookup the state we must explicitly skip it in order not to invoke it
  twice.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 kernel/cpu.c | 162 +++++++++++++++++++++++++++++------------------------------
 1 file changed, 80 insertions(+), 82 deletions(-)

diff --git a/kernel/cpu.c b/kernel/cpu.c
index 341bf80f80bd..3f66f87246be 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -37,8 +37,9 @@
  * @thread:	Pointer to the hotplug thread
  * @should_run:	Thread should execute
  * @rollback:	Perform a rollback
- * @cb_stat:	The state for a single callback (install/uninstall)
- * @cb:		Single callback function (install/uninstall)
+ * @single:	Single callback invocation
+ * @bringup:	Single callback bringup or teardown selector
+ * @cb_state:	The state for a single callback (install/uninstall)
  * @result:	Result of the operation
  * @done:	Signal completion to the issuer of the task
  */
@@ -49,8 +50,9 @@ struct cpuhp_cpu_state {
 	struct task_struct	*thread;
 	bool			should_run;
 	bool			rollback;
+	bool			single;
+	bool			bringup;
 	enum cpuhp_state	cb_state;
-	int			(*cb)(unsigned int cpu);
 	int			result;
 	struct completion	done;
 #endif
@@ -79,24 +81,43 @@ static DEFINE_MUTEX(cpuhp_state_mutex);
 static struct cpuhp_step cpuhp_bp_states[];
 static struct cpuhp_step cpuhp_ap_states[];
 
+static bool cpuhp_is_ap_state(enum cpuhp_state state)
+{
+	/*
+	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
+	 * purposes as that state is handled explicitly in cpu_down.
+	 */
+	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
+}
+
+static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
+{
+	struct cpuhp_step *sp;
+
+	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
+	return sp + state;
+}
+
 /**
  * cpuhp_invoke_callback _ Invoke the callbacks for a given state
  * @cpu:	The cpu for which the callback should be invoked
  * @step:	The step in the state machine
- * @cb:		The callback function to invoke
+ * @bringup:	True if the bringup callback should be invoked
  *
  * Called from cpu hotplug and from the state register machinery
  */
-static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state step,
-				 int (*cb)(unsigned int))
+static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
+				 bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+	struct cpuhp_step *step = cpuhp_get_step(state);
+	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
 	int ret = 0;
 
 	if (cb) {
-		trace_cpuhp_enter(cpu, st->target, step, cb);
+		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
-		trace_cpuhp_exit(cpu, st->state, step, ret);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
 	}
 	return ret;
 }
@@ -363,62 +384,55 @@ static int bringup_cpu(unsigned int cpu)
 /*
  * Hotplug state machine related functions
  */
-static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st,
-			  struct cpuhp_step *steps)
+static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state++; st->state < st->target; st->state++) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->startup);
+			cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
 static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-				struct cpuhp_step *steps, enum cpuhp_state target)
+				enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
-
-		ret = cpuhp_invoke_callback(cpu, st->state, step->teardown);
+		ret = cpuhp_invoke_callback(cpu, st->state, false);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_down(cpu, st, steps);
+			undo_cpu_down(cpu, st);
 			break;
 		}
 	}
 	return ret;
 }
 
-static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st,
-			struct cpuhp_step *steps)
+static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state--; st->state > st->target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->teardown);
+			cpuhp_invoke_callback(cpu, st->state, false);
 	}
 }
 
 static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-			      struct cpuhp_step *steps, enum cpuhp_state target)
+			      enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = steps + st->state;
-		ret = cpuhp_invoke_callback(cpu, st->state, step->startup);
+		ret = cpuhp_invoke_callback(cpu, st->state, true);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_up(cpu, st, steps);
+			undo_cpu_up(cpu, st);
 			break;
 		}
 	}
@@ -447,13 +461,13 @@ static int cpuhp_ap_offline(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	enum cpuhp_state target = max((int)st->target, CPUHP_TEARDOWN_CPU);
 
-	return cpuhp_down_callbacks(cpu, st, cpuhp_ap_states, target);
+	return cpuhp_down_callbacks(cpu, st, target);
 }
 
 /* Execute the online startup callbacks. Used to be CPU_ONLINE */
 static int cpuhp_ap_online(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
-	return cpuhp_up_callbacks(cpu, st, cpuhp_ap_states, st->target);
+	return cpuhp_up_callbacks(cpu, st, st->target);
 }
 
 /*
@@ -476,18 +490,20 @@ static void cpuhp_thread_fun(unsigned int cpu)
 	st->should_run = false;
 
 	/* Single callback invocation for [un]install ? */
-	if (st->cb) {
+	if (st->single) {
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 			local_irq_enable();
 		} else {
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
 
-		undo_cpu_down(cpu, st, cpuhp_ap_states);
+		undo_cpu_down(cpu, st);
 		/*
 		 * This is a momentary workaround to keep the notifier users
 		 * happy. Will go away once we got rid of the notifiers.
@@ -509,8 +525,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 }
 
 /* Invoke a single callback on a remote cpu */
-static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
-				    int (*cb)(unsigned int))
+static int
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -522,10 +538,12 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, cb);
+		return cpuhp_invoke_callback(cpu, state, bringup);
 
 	st->cb_state = state;
-	st->cb = cb;
+	st->single = true;
+	st->bringup = bringup;
+
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -541,7 +559,7 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 static void __cpuhp_kick_ap_work(struct cpuhp_cpu_state *st)
 {
 	st->result = 0;
-	st->cb = NULL;
+	st->single = false;
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -692,12 +710,16 @@ static int take_cpu_down(void *_param)
 	if (err < 0)
 		return err;
 
+	/*
+	 * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not
+	 * do this step again.
+	 */
+	WARN_ON(st->state != CPUHP_TEARDOWN_CPU);
+	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
-	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = cpuhp_ap_states + st->state;
+	for (; st->state > target; st->state--)
+		cpuhp_invoke_callback(cpu, st->state, false);
 
-		cpuhp_invoke_callback(cpu, st->state, step->teardown);
-	}
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
 	/* Park the stopper thread */
@@ -836,7 +858,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
 	 * The AP brought itself down to CPUHP_TEARDOWN_CPU. So we need
 	 * to do the further cleanups.
 	 */
-	ret = cpuhp_down_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_down_callbacks(cpu, st, target);
 	if (ret && st->state > CPUHP_TEARDOWN_CPU && st->state < prev_state) {
 		st->target = prev_state;
 		st->rollback = true;
@@ -890,11 +912,8 @@ void notify_cpu_starting(unsigned int cpu)
 	enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE);
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = cpuhp_ap_states + st->state;
-		cpuhp_invoke_callback(cpu, st->state, step->startup);
+		cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
@@ -979,7 +998,7 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
 	 * responsible for bringing it up to the target state.
 	 */
 	target = min((int)target, CPUHP_BRINGUP_CPU);
-	ret = cpuhp_up_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_up_callbacks(cpu, st, target);
 out:
 	cpu_hotplug_done();
 	return ret;
@@ -1356,23 +1375,6 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 	return 0;
 }
 
-static bool cpuhp_is_ap_state(enum cpuhp_state state)
-{
-	/*
-	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
-	 * purposes as that state is handled explicitely in cpu_down.
-	 */
-	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
-}
-
-static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
-{
-	struct cpuhp_step *sp;
-
-	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
-	return sp + state;
-}
-
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
@@ -1398,12 +1400,12 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
-			    int (*cb)(unsigned int), bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
 
-	if (!cb)
+	if ((bringup && !sp->startup) || (!bringup && !sp->teardown))
 		return 0;
 	/*
 	 * The non AP bound callbacks can fail on bringup. On teardown
@@ -1411,11 +1413,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, cb);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, cb);
+		ret = cpuhp_invoke_callback(cpu, state, bringup);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, cb);
+	ret = cpuhp_invoke_callback(cpu, state, bringup);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1426,14 +1428,10 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
-				   int (*teardown)(unsigned int cpu))
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 {
 	int cpu;
 
-	if (!teardown)
-		return;
-
 	/* Roll back the already executed steps on the other cpus */
 	for_each_present_cpu(cpu) {
 		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
@@ -1444,7 +1442,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 }
 
@@ -1519,9 +1517,10 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, startup, true);
+		ret = cpuhp_issue_call(cpu, state, true);
 		if (ret) {
-			cpuhp_rollback_install(cpu, state, teardown);
+			if (teardown)
+				cpuhp_rollback_install(cpu, state);
 			cpuhp_store_callbacks(state, NULL, NULL, NULL);
 			goto out;
 		}
@@ -1545,14 +1544,13 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
-	int (*teardown)(unsigned int cpu) = cpuhp_get_teardown_cb(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
-	if (!invoke || !teardown)
+	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
 	/*
@@ -1565,7 +1563,7 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 remove:
 	cpuhp_store_callbacks(state, NULL, NULL, NULL);
-- 
2.8.1

[tip:smp/hotplug] cpu/hotplug: Rework callback invocation logic

[tip-bot for Thomas Gleixner]

Commit-ID:  8e8afc3614b3f9d6c1495d324a5e1683950c28a6
Gitweb:     http://git.kernel.org/tip/8e8afc3614b3f9d6c1495d324a5e1683950c28a6
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:38 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:28 +0200

cpu/hotplug: Rework callback invocation logic

This is preparation for the following patch.
This rework here changes the arguments of cpuhp_invoke_callback(). It
passes now `state' and whether `startup' or `teardown' callback should
be invoked. The callback then is looked up by the function.

The following is a clanup of callers:
- cpuhp_issue_call() has one argument less
- struct cpuhp_cpu_state (which is used by the hotplug thread) gets also
  its callback removed. The decision if it is a single callback
  invocation moved to the `single' variable. Also a `bringup' variable
  has been added to distinguish between startup and teardown callback.
- take_cpu_down() needs to start one step earlier. We always get here
  via CPUHP_TEARDOWN_CPU callback. Before that change cpuhp_ap_states +
  CPUHP_TEARDOWN_CPU pointed to an empty entry because TEARDOWN is saved
  in bp_states for this reason. Now that we use cpuhp_get_step() to
  lookup the state we must explicitly skip it in order not to invoke it
  twice.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
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/1471024183-12666-2-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 kernel/cpu.c | 162 +++++++++++++++++++++++++++++------------------------------
 1 file changed, 80 insertions(+), 82 deletions(-)

diff --git a/kernel/cpu.c b/kernel/cpu.c
index ec12b72..d36d8e0 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -37,8 +37,9 @@
  * @thread:	Pointer to the hotplug thread
  * @should_run:	Thread should execute
  * @rollback:	Perform a rollback
- * @cb_stat:	The state for a single callback (install/uninstall)
- * @cb:		Single callback function (install/uninstall)
+ * @single:	Single callback invocation
+ * @bringup:	Single callback bringup or teardown selector
+ * @cb_state:	The state for a single callback (install/uninstall)
  * @result:	Result of the operation
  * @done:	Signal completion to the issuer of the task
  */
@@ -49,8 +50,9 @@ struct cpuhp_cpu_state {
 	struct task_struct	*thread;
 	bool			should_run;
 	bool			rollback;
+	bool			single;
+	bool			bringup;
 	enum cpuhp_state	cb_state;
-	int			(*cb)(unsigned int cpu);
 	int			result;
 	struct completion	done;
 #endif
@@ -79,24 +81,43 @@ static DEFINE_MUTEX(cpuhp_state_mutex);
 static struct cpuhp_step cpuhp_bp_states[];
 static struct cpuhp_step cpuhp_ap_states[];
 
+static bool cpuhp_is_ap_state(enum cpuhp_state state)
+{
+	/*
+	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
+	 * purposes as that state is handled explicitly in cpu_down.
+	 */
+	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
+}
+
+static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
+{
+	struct cpuhp_step *sp;
+
+	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
+	return sp + state;
+}
+
 /**
  * cpuhp_invoke_callback _ Invoke the callbacks for a given state
  * @cpu:	The cpu for which the callback should be invoked
  * @step:	The step in the state machine
- * @cb:		The callback function to invoke
+ * @bringup:	True if the bringup callback should be invoked
  *
  * Called from cpu hotplug and from the state register machinery
  */
-static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state step,
-				 int (*cb)(unsigned int))
+static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
+				 bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+	struct cpuhp_step *step = cpuhp_get_step(state);
+	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
 	int ret = 0;
 
 	if (cb) {
-		trace_cpuhp_enter(cpu, st->target, step, cb);
+		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
-		trace_cpuhp_exit(cpu, st->state, step, ret);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
 	}
 	return ret;
 }
@@ -371,62 +392,55 @@ static int bringup_cpu(unsigned int cpu)
 /*
  * Hotplug state machine related functions
  */
-static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st,
-			  struct cpuhp_step *steps)
+static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state++; st->state < st->target; st->state++) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->startup);
+			cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
 static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-				struct cpuhp_step *steps, enum cpuhp_state target)
+				enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
-
-		ret = cpuhp_invoke_callback(cpu, st->state, step->teardown);
+		ret = cpuhp_invoke_callback(cpu, st->state, false);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_down(cpu, st, steps);
+			undo_cpu_down(cpu, st);
 			break;
 		}
 	}
 	return ret;
 }
 
-static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st,
-			struct cpuhp_step *steps)
+static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state--; st->state > st->target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->teardown);
+			cpuhp_invoke_callback(cpu, st->state, false);
 	}
 }
 
 static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-			      struct cpuhp_step *steps, enum cpuhp_state target)
+			      enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = steps + st->state;
-		ret = cpuhp_invoke_callback(cpu, st->state, step->startup);
+		ret = cpuhp_invoke_callback(cpu, st->state, true);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_up(cpu, st, steps);
+			undo_cpu_up(cpu, st);
 			break;
 		}
 	}
@@ -455,13 +469,13 @@ static int cpuhp_ap_offline(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	enum cpuhp_state target = max((int)st->target, CPUHP_TEARDOWN_CPU);
 
-	return cpuhp_down_callbacks(cpu, st, cpuhp_ap_states, target);
+	return cpuhp_down_callbacks(cpu, st, target);
 }
 
 /* Execute the online startup callbacks. Used to be CPU_ONLINE */
 static int cpuhp_ap_online(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
-	return cpuhp_up_callbacks(cpu, st, cpuhp_ap_states, st->target);
+	return cpuhp_up_callbacks(cpu, st, st->target);
 }
 
 /*
@@ -484,18 +498,20 @@ static void cpuhp_thread_fun(unsigned int cpu)
 	st->should_run = false;
 
 	/* Single callback invocation for [un]install ? */
-	if (st->cb) {
+	if (st->single) {
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 			local_irq_enable();
 		} else {
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
 
-		undo_cpu_down(cpu, st, cpuhp_ap_states);
+		undo_cpu_down(cpu, st);
 		/*
 		 * This is a momentary workaround to keep the notifier users
 		 * happy. Will go away once we got rid of the notifiers.
@@ -517,8 +533,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 }
 
 /* Invoke a single callback on a remote cpu */
-static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
-				    int (*cb)(unsigned int))
+static int
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -530,10 +546,12 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, cb);
+		return cpuhp_invoke_callback(cpu, state, bringup);
 
 	st->cb_state = state;
-	st->cb = cb;
+	st->single = true;
+	st->bringup = bringup;
+
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -549,7 +567,7 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 static void __cpuhp_kick_ap_work(struct cpuhp_cpu_state *st)
 {
 	st->result = 0;
-	st->cb = NULL;
+	st->single = false;
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -700,12 +718,16 @@ static int take_cpu_down(void *_param)
 	if (err < 0)
 		return err;
 
+	/*
+	 * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not
+	 * do this step again.
+	 */
+	WARN_ON(st->state != CPUHP_TEARDOWN_CPU);
+	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
-	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = cpuhp_ap_states + st->state;
+	for (; st->state > target; st->state--)
+		cpuhp_invoke_callback(cpu, st->state, false);
 
-		cpuhp_invoke_callback(cpu, st->state, step->teardown);
-	}
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
 	/* Park the stopper thread */
@@ -844,7 +866,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
 	 * The AP brought itself down to CPUHP_TEARDOWN_CPU. So we need
 	 * to do the further cleanups.
 	 */
-	ret = cpuhp_down_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_down_callbacks(cpu, st, target);
 	if (ret && st->state > CPUHP_TEARDOWN_CPU && st->state < prev_state) {
 		st->target = prev_state;
 		st->rollback = true;
@@ -898,11 +920,8 @@ void notify_cpu_starting(unsigned int cpu)
 	enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE);
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = cpuhp_ap_states + st->state;
-		cpuhp_invoke_callback(cpu, st->state, step->startup);
+		cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
@@ -987,7 +1006,7 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
 	 * responsible for bringing it up to the target state.
 	 */
 	target = min((int)target, CPUHP_BRINGUP_CPU);
-	ret = cpuhp_up_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_up_callbacks(cpu, st, target);
 out:
 	cpu_hotplug_done();
 	return ret;
@@ -1364,23 +1383,6 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 	return 0;
 }
 
-static bool cpuhp_is_ap_state(enum cpuhp_state state)
-{
-	/*
-	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
-	 * purposes as that state is handled explicitely in cpu_down.
-	 */
-	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
-}
-
-static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
-{
-	struct cpuhp_step *sp;
-
-	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
-	return sp + state;
-}
-
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
@@ -1406,12 +1408,12 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
-			    int (*cb)(unsigned int), bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
 
-	if (!cb)
+	if ((bringup && !sp->startup) || (!bringup && !sp->teardown))
 		return 0;
 	/*
 	 * The non AP bound callbacks can fail on bringup. On teardown
@@ -1419,11 +1421,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, cb);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, cb);
+		ret = cpuhp_invoke_callback(cpu, state, bringup);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, cb);
+	ret = cpuhp_invoke_callback(cpu, state, bringup);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1434,14 +1436,10 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
-				   int (*teardown)(unsigned int cpu))
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 {
 	int cpu;
 
-	if (!teardown)
-		return;
-
 	/* Roll back the already executed steps on the other cpus */
 	for_each_present_cpu(cpu) {
 		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
@@ -1452,7 +1450,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 }
 
@@ -1527,9 +1525,10 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, startup, true);
+		ret = cpuhp_issue_call(cpu, state, true);
 		if (ret) {
-			cpuhp_rollback_install(cpu, state, teardown);
+			if (teardown)
+				cpuhp_rollback_install(cpu, state);
 			cpuhp_store_callbacks(state, NULL, NULL, NULL);
 			goto out;
 		}
@@ -1553,14 +1552,13 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
-	int (*teardown)(unsigned int cpu) = cpuhp_get_teardown_cb(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
-	if (!invoke || !teardown)
+	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
 	/*
@@ -1573,7 +1571,7 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 remove:
 	cpuhp_store_callbacks(state, NULL, NULL, NULL);

[tip:smp/hotplug] cpu/hotplug: Rework callback invocation logic

[tip-bot for Thomas Gleixner]

Commit-ID:  a724632ca0c84b494875e9367e07e29472c139ba
Gitweb:     http://git.kernel.org/tip/a724632ca0c84b494875e9367e07e29472c139ba
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:38 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:05 +0200

cpu/hotplug: Rework callback invocation logic

This is preparation for the following patch.
This rework here changes the arguments of cpuhp_invoke_callback(). It
passes now `state' and whether `startup' or `teardown' callback should
be invoked. The callback then is looked up by the function.

The following is a clanup of callers:
- cpuhp_issue_call() has one argument less
- struct cpuhp_cpu_state (which is used by the hotplug thread) gets also
  its callback removed. The decision if it is a single callback
  invocation moved to the `single' variable. Also a `bringup' variable
  has been added to distinguish between startup and teardown callback.
- take_cpu_down() needs to start one step earlier. We always get here
  via CPUHP_TEARDOWN_CPU callback. Before that change cpuhp_ap_states +
  CPUHP_TEARDOWN_CPU pointed to an empty entry because TEARDOWN is saved
  in bp_states for this reason. Now that we use cpuhp_get_step() to
  lookup the state we must explicitly skip it in order not to invoke it
  twice.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
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/1471024183-12666-2-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 kernel/cpu.c | 162 +++++++++++++++++++++++++++++------------------------------
 1 file changed, 80 insertions(+), 82 deletions(-)

diff --git a/kernel/cpu.c b/kernel/cpu.c
index ec12b72..d36d8e0 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -37,8 +37,9 @@
  * @thread:	Pointer to the hotplug thread
  * @should_run:	Thread should execute
  * @rollback:	Perform a rollback
- * @cb_stat:	The state for a single callback (install/uninstall)
- * @cb:		Single callback function (install/uninstall)
+ * @single:	Single callback invocation
+ * @bringup:	Single callback bringup or teardown selector
+ * @cb_state:	The state for a single callback (install/uninstall)
  * @result:	Result of the operation
  * @done:	Signal completion to the issuer of the task
  */
@@ -49,8 +50,9 @@ struct cpuhp_cpu_state {
 	struct task_struct	*thread;
 	bool			should_run;
 	bool			rollback;
+	bool			single;
+	bool			bringup;
 	enum cpuhp_state	cb_state;
-	int			(*cb)(unsigned int cpu);
 	int			result;
 	struct completion	done;
 #endif
@@ -79,24 +81,43 @@ static DEFINE_MUTEX(cpuhp_state_mutex);
 static struct cpuhp_step cpuhp_bp_states[];
 static struct cpuhp_step cpuhp_ap_states[];
 
+static bool cpuhp_is_ap_state(enum cpuhp_state state)
+{
+	/*
+	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
+	 * purposes as that state is handled explicitly in cpu_down.
+	 */
+	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
+}
+
+static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
+{
+	struct cpuhp_step *sp;
+
+	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
+	return sp + state;
+}
+
 /**
  * cpuhp_invoke_callback _ Invoke the callbacks for a given state
  * @cpu:	The cpu for which the callback should be invoked
  * @step:	The step in the state machine
- * @cb:		The callback function to invoke
+ * @bringup:	True if the bringup callback should be invoked
  *
  * Called from cpu hotplug and from the state register machinery
  */
-static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state step,
-				 int (*cb)(unsigned int))
+static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
+				 bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+	struct cpuhp_step *step = cpuhp_get_step(state);
+	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
 	int ret = 0;
 
 	if (cb) {
-		trace_cpuhp_enter(cpu, st->target, step, cb);
+		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
-		trace_cpuhp_exit(cpu, st->state, step, ret);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
 	}
 	return ret;
 }
@@ -371,62 +392,55 @@ static int bringup_cpu(unsigned int cpu)
 /*
  * Hotplug state machine related functions
  */
-static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st,
-			  struct cpuhp_step *steps)
+static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state++; st->state < st->target; st->state++) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->startup);
+			cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
 static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-				struct cpuhp_step *steps, enum cpuhp_state target)
+				enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
-
-		ret = cpuhp_invoke_callback(cpu, st->state, step->teardown);
+		ret = cpuhp_invoke_callback(cpu, st->state, false);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_down(cpu, st, steps);
+			undo_cpu_down(cpu, st);
 			break;
 		}
 	}
 	return ret;
 }
 
-static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st,
-			struct cpuhp_step *steps)
+static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	for (st->state--; st->state > st->target; st->state--) {
-		struct cpuhp_step *step = steps + st->state;
+		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, step->teardown);
+			cpuhp_invoke_callback(cpu, st->state, false);
 	}
 }
 
 static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
-			      struct cpuhp_step *steps, enum cpuhp_state target)
+			      enum cpuhp_state target)
 {
 	enum cpuhp_state prev_state = st->state;
 	int ret = 0;
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = steps + st->state;
-		ret = cpuhp_invoke_callback(cpu, st->state, step->startup);
+		ret = cpuhp_invoke_callback(cpu, st->state, true);
 		if (ret) {
 			st->target = prev_state;
-			undo_cpu_up(cpu, st, steps);
+			undo_cpu_up(cpu, st);
 			break;
 		}
 	}
@@ -455,13 +469,13 @@ static int cpuhp_ap_offline(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
 	enum cpuhp_state target = max((int)st->target, CPUHP_TEARDOWN_CPU);
 
-	return cpuhp_down_callbacks(cpu, st, cpuhp_ap_states, target);
+	return cpuhp_down_callbacks(cpu, st, target);
 }
 
 /* Execute the online startup callbacks. Used to be CPU_ONLINE */
 static int cpuhp_ap_online(unsigned int cpu, struct cpuhp_cpu_state *st)
 {
-	return cpuhp_up_callbacks(cpu, st, cpuhp_ap_states, st->target);
+	return cpuhp_up_callbacks(cpu, st, st->target);
 }
 
 /*
@@ -484,18 +498,20 @@ static void cpuhp_thread_fun(unsigned int cpu)
 	st->should_run = false;
 
 	/* Single callback invocation for [un]install ? */
-	if (st->cb) {
+	if (st->single) {
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 			local_irq_enable();
 		} else {
-			ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb);
+			ret = cpuhp_invoke_callback(cpu, st->cb_state,
+						    st->bringup);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
 
-		undo_cpu_down(cpu, st, cpuhp_ap_states);
+		undo_cpu_down(cpu, st);
 		/*
 		 * This is a momentary workaround to keep the notifier users
 		 * happy. Will go away once we got rid of the notifiers.
@@ -517,8 +533,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 }
 
 /* Invoke a single callback on a remote cpu */
-static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
-				    int (*cb)(unsigned int))
+static int
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -530,10 +546,12 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, cb);
+		return cpuhp_invoke_callback(cpu, state, bringup);
 
 	st->cb_state = state;
-	st->cb = cb;
+	st->single = true;
+	st->bringup = bringup;
+
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -549,7 +567,7 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state,
 static void __cpuhp_kick_ap_work(struct cpuhp_cpu_state *st)
 {
 	st->result = 0;
-	st->cb = NULL;
+	st->single = false;
 	/*
 	 * Make sure the above stores are visible before should_run becomes
 	 * true. Paired with the mb() above in cpuhp_thread_fun()
@@ -700,12 +718,16 @@ static int take_cpu_down(void *_param)
 	if (err < 0)
 		return err;
 
+	/*
+	 * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not
+	 * do this step again.
+	 */
+	WARN_ON(st->state != CPUHP_TEARDOWN_CPU);
+	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
-	for (; st->state > target; st->state--) {
-		struct cpuhp_step *step = cpuhp_ap_states + st->state;
+	for (; st->state > target; st->state--)
+		cpuhp_invoke_callback(cpu, st->state, false);
 
-		cpuhp_invoke_callback(cpu, st->state, step->teardown);
-	}
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
 	/* Park the stopper thread */
@@ -844,7 +866,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
 	 * The AP brought itself down to CPUHP_TEARDOWN_CPU. So we need
 	 * to do the further cleanups.
 	 */
-	ret = cpuhp_down_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_down_callbacks(cpu, st, target);
 	if (ret && st->state > CPUHP_TEARDOWN_CPU && st->state < prev_state) {
 		st->target = prev_state;
 		st->rollback = true;
@@ -898,11 +920,8 @@ void notify_cpu_starting(unsigned int cpu)
 	enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE);
 
 	while (st->state < target) {
-		struct cpuhp_step *step;
-
 		st->state++;
-		step = cpuhp_ap_states + st->state;
-		cpuhp_invoke_callback(cpu, st->state, step->startup);
+		cpuhp_invoke_callback(cpu, st->state, true);
 	}
 }
 
@@ -987,7 +1006,7 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
 	 * responsible for bringing it up to the target state.
 	 */
 	target = min((int)target, CPUHP_BRINGUP_CPU);
-	ret = cpuhp_up_callbacks(cpu, st, cpuhp_bp_states, target);
+	ret = cpuhp_up_callbacks(cpu, st, target);
 out:
 	cpu_hotplug_done();
 	return ret;
@@ -1364,23 +1383,6 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 	return 0;
 }
 
-static bool cpuhp_is_ap_state(enum cpuhp_state state)
-{
-	/*
-	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
-	 * purposes as that state is handled explicitely in cpu_down.
-	 */
-	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
-}
-
-static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
-{
-	struct cpuhp_step *sp;
-
-	sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states;
-	return sp + state;
-}
-
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
@@ -1406,12 +1408,12 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
-			    int (*cb)(unsigned int), bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
 
-	if (!cb)
+	if ((bringup && !sp->startup) || (!bringup && !sp->teardown))
 		return 0;
 	/*
 	 * The non AP bound callbacks can fail on bringup. On teardown
@@ -1419,11 +1421,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, cb);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, cb);
+		ret = cpuhp_invoke_callback(cpu, state, bringup);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, cb);
+	ret = cpuhp_invoke_callback(cpu, state, bringup);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1434,14 +1436,10 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state,
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
-				   int (*teardown)(unsigned int cpu))
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 {
 	int cpu;
 
-	if (!teardown)
-		return;
-
 	/* Roll back the already executed steps on the other cpus */
 	for_each_present_cpu(cpu) {
 		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
@@ -1452,7 +1450,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 }
 
@@ -1527,9 +1525,10 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, startup, true);
+		ret = cpuhp_issue_call(cpu, state, true);
 		if (ret) {
-			cpuhp_rollback_install(cpu, state, teardown);
+			if (teardown)
+				cpuhp_rollback_install(cpu, state);
 			cpuhp_store_callbacks(state, NULL, NULL, NULL);
 			goto out;
 		}
@@ -1553,14 +1552,13 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
-	int (*teardown)(unsigned int cpu) = cpuhp_get_teardown_cb(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
-	if (!invoke || !teardown)
+	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
 	/*
@@ -1573,7 +1571,7 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, teardown, false);
+			cpuhp_issue_call(cpu, state, false);
 	}
 remove:
 	cpuhp_store_callbacks(state, NULL, NULL, NULL);

[PATCH 2/6] cpu/hotplug: Add multi instance support

[Sebastian Andrzej Siewior]

From: Thomas Gleixner <tglx@linutronix.de>

This patch adds the ability for a given state to have multiple
instances. Until now all states have a single instance and the startup /
teardown callback use global variables.
A few drivers need to perform a the same callbacks on multiple
"instances". Currently we have three drivers in tree which all have a
global list which they iterate over. With multi instance they support
don't need their private list and the functionality has been moved into
core code. Plus we hold the hotplug lock in core so no cpus comes/goes
while instances are registered and we do rollback in error case :)

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 include/linux/cpuhotplug.h   |  97 ++++++++++++++++++-
 include/trace/events/cpuhp.h |  28 ++++++
 kernel/cpu.c                 | 215 ++++++++++++++++++++++++++++++++++++-------
 3 files changed, 303 insertions(+), 37 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 242bf530edfc..81cc20789a3e 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -99,7 +99,7 @@ enum cpuhp_state {
 
 int __cpuhp_setup_state(enum cpuhp_state state,	const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu));
+			int (*teardown)(unsigned int cpu), bool multi_instance);
 
 /**
  * cpuhp_setup_state - Setup hotplug state callbacks with calling the callbacks
@@ -116,7 +116,7 @@ static inline int cpuhp_setup_state(enum cpuhp_state state,
 				    int (*startup)(unsigned int cpu),
 				    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, true, startup, teardown);
+	return __cpuhp_setup_state(state, name, true, startup, teardown, false);
 }
 
 /**
@@ -135,7 +135,66 @@ static inline int cpuhp_setup_state_nocalls(enum cpuhp_state state,
 					    int (*startup)(unsigned int cpu),
 					    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, false, startup, teardown);
+	return __cpuhp_setup_state(state, name, false, startup, teardown,
+				   false);
+}
+
+/**
+ * cpuhp_setup_state_multi - Add callbacks for multi state
+ * @state:	The state for which the calls are installed
+ * @name:	Name of the callback.
+ * @startup:	startup callback function
+ * @teardown:	teardown callback function
+ *
+ * Sets the internal multi_instance flag and prepares a state to work as a multi
+ * instance callback. No callbacks are invoked at this point. The callbacks are
+ * invoked once an instance for this state are registered via
+ * @cpuhp_state_add_instance or @cpuhp_state_add_instance_nocalls.
+ */
+static inline int cpuhp_setup_state_multi(enum cpuhp_state state,
+					  const char *name,
+					  int (*startup)(unsigned int cpu,
+							 struct hlist_node *node),
+					  int (*teardown)(unsigned int cpu,
+							  struct hlist_node *node))
+{
+	return __cpuhp_setup_state(state, name, false,
+				   (void *) startup,
+				   (void *) teardown, true);
+}
+
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke);
+
+/**
+ * cpuhp_state_add_instance - Add an instance for a state and invoke startup
+ *                            callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state and invokes the startup callback on
+ * the present cpus which have already reached the @state. The @state must have
+ * been earlier marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance(enum cpuhp_state state,
+					   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_add_instance_nocalls - Add an instance for a state without
+ *                                    invoking the startup callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state The @state must have been earlier
+ * marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance_nocalls(enum cpuhp_state state,
+						   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, false);
 }
 
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke);
@@ -162,6 +221,38 @@ static inline void cpuhp_remove_state_nocalls(enum cpuhp_state state)
 	__cpuhp_remove_state(state, false);
 }
 
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke);
+
+/**
+ * cpuhp_state_remove_instance - Remove hotplug instance from state and invoke
+ *                               the teardown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance and invokes the teardown callback on the present cpus
+ * which have already reached the @state.
+ */
+static inline int cpuhp_state_remove_instance(enum cpuhp_state state,
+					      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_remove_instance_nocalls - Remove hotplug instance from state
+ *					 without invoking the reatdown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance without invoking the teardown callback.
+ */
+static inline int cpuhp_state_remove_instance_nocalls(enum cpuhp_state state,
+						      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, false);
+}
+
 #ifdef CONFIG_SMP
 void cpuhp_online_idle(enum cpuhp_state state);
 #else
diff --git a/include/trace/events/cpuhp.h b/include/trace/events/cpuhp.h
index a72bd93ec7e5..996953db91d7 100644
--- a/include/trace/events/cpuhp.h
+++ b/include/trace/events/cpuhp.h
@@ -33,6 +33,34 @@ TRACE_EVENT(cpuhp_enter,
 		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
 );
 
+TRACE_EVENT(cpuhp_multi_enter,
+
+	TP_PROTO(unsigned int cpu,
+		 int target,
+		 int idx,
+		 int (*fun)(unsigned int, struct hlist_node *),
+		 struct hlist_node *node),
+
+	TP_ARGS(cpu, target, idx, fun, node),
+
+	TP_STRUCT__entry(
+		__field( unsigned int,	cpu		)
+		__field( int,		target		)
+		__field( int,		idx		)
+		__field( void *,	fun		)
+	),
+
+	TP_fast_assign(
+		__entry->cpu	= cpu;
+		__entry->target	= target;
+		__entry->idx	= idx;
+		__entry->fun	= fun;
+	),
+
+	TP_printk("cpu: %04u target: %3d step: %3d (%pf)",
+		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
+);
+
 TRACE_EVENT(cpuhp_exit,
 
 	TP_PROTO(unsigned int cpu,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 3f66f87246be..1b3e47464cc7 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -52,6 +52,7 @@ struct cpuhp_cpu_state {
 	bool			rollback;
 	bool			single;
 	bool			bringup;
+	struct hlist_node	*node;
 	enum cpuhp_state	cb_state;
 	int			result;
 	struct completion	done;
@@ -70,11 +71,21 @@ static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state);
  * @cant_stop:	Bringup/teardown can't be stopped at this step
  */
 struct cpuhp_step {
-	const char	*name;
-	int		(*startup)(unsigned int cpu);
-	int		(*teardown)(unsigned int cpu);
-	bool		skip_onerr;
-	bool		cant_stop;
+	const char		*name;
+	union {
+		int		(*startup)(unsigned int cpu);
+		int		(*startup_multi)(unsigned int cpu,
+						 struct hlist_node *node);
+	};
+	union {
+		int		(*teardown)(unsigned int cpu);
+		int		(*teardown_multi)(unsigned int cpu,
+						  struct hlist_node *node);
+	};
+	struct hlist_head	list;
+	bool			skip_onerr;
+	bool			cant_stop;
+	bool			multi_instance;
 };
 
 static DEFINE_MUTEX(cpuhp_state_mutex);
@@ -104,20 +115,59 @@ static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
  * @step:	The step in the state machine
  * @bringup:	True if the bringup callback should be invoked
  *
- * Called from cpu hotplug and from the state register machinery
+ * Called from cpu hotplug and from the state register machinery.
  */
 static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
-				 bool bringup)
+				 bool bringup, struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 	struct cpuhp_step *step = cpuhp_get_step(state);
-	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
-	int ret = 0;
+	int (*cbm)(unsigned int cpu, struct hlist_node *node);
+	int (*cb)(unsigned int cpu);
+	int ret, cnt;
 
-	if (cb) {
+	if (!step->multi_instance) {
+		cb = bringup ? step->startup : step->teardown;
+		if (!cb)
+			return 0;
 		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
 		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+	cbm = bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return 0;
+
+	/* Single invocation for instance add/remove */
+	if (node) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+
+	/* State transition. Invoke on all instances */
+	cnt = 0;
+	hlist_for_each(node, &step->list) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		if (ret)
+			goto err;
+		cnt++;
+	}
+	return 0;
+err:
+	/* Rollback the instances if one failed */
+	cbm = !bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return ret;
+
+	hlist_for_each(node, &step->list) {
+		if (!cnt--)
+			break;
+		cbm(cpu, node);
 	}
 	return ret;
 }
@@ -390,7 +440,7 @@ static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, true);
+			cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -401,7 +451,7 @@ static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		ret = cpuhp_invoke_callback(cpu, st->state, false);
+		ret = cpuhp_invoke_callback(cpu, st->state, false, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_down(cpu, st);
@@ -417,7 +467,7 @@ static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, false);
+			cpuhp_invoke_callback(cpu, st->state, false, NULL);
 	}
 }
 
@@ -429,7 +479,7 @@ static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 
 	while (st->state < target) {
 		st->state++;
-		ret = cpuhp_invoke_callback(cpu, st->state, true);
+		ret = cpuhp_invoke_callback(cpu, st->state, true, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_up(cpu, st);
@@ -494,11 +544,11 @@ static void cpuhp_thread_fun(unsigned int cpu)
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 			local_irq_enable();
 		} else {
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
@@ -526,7 +576,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 
 /* Invoke a single callback on a remote cpu */
 static int
-cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup,
+			 struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -538,11 +589,12 @@ cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, bringup);
+		return cpuhp_invoke_callback(cpu, state, bringup, node);
 
 	st->cb_state = state;
 	st->single = true;
 	st->bringup = bringup;
+	st->node = node;
 
 	/*
 	 * Make sure the above stores are visible before should_run becomes
@@ -718,7 +770,7 @@ static int take_cpu_down(void *_param)
 	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
 	for (; st->state > target; st->state--)
-		cpuhp_invoke_callback(cpu, st->state, false);
+		cpuhp_invoke_callback(cpu, st->state, false, NULL);
 
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
@@ -913,7 +965,7 @@ void notify_cpu_starting(unsigned int cpu)
 
 	while (st->state < target) {
 		st->state++;
-		cpuhp_invoke_callback(cpu, st->state, true);
+		cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -1378,7 +1430,8 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
-				  int (*teardown)(unsigned int cpu))
+				  int (*teardown)(unsigned int cpu),
+				  bool multi_instance)
 {
 	/* (Un)Install the callbacks for further cpu hotplug operations */
 	struct cpuhp_step *sp;
@@ -1388,6 +1441,8 @@ static void cpuhp_store_callbacks(enum cpuhp_state state,
 	sp->startup = startup;
 	sp->teardown = teardown;
 	sp->name = name;
+	sp->multi_instance = multi_instance;
+	INIT_HLIST_HEAD(&sp->list);
 	mutex_unlock(&cpuhp_state_mutex);
 }
 
@@ -1400,7 +1455,8 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup,
+			    struct hlist_node *node)
 {
 	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
@@ -1413,11 +1469,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup, node);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, bringup);
+	ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1428,7 +1484,8 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
+				   struct hlist_node *node)
 {
 	int cpu;
 
@@ -1442,7 +1499,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, node);
 	}
 }
 
@@ -1469,6 +1526,52 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 	return -ENOSPC;
 }
 
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke)
+{
+	struct cpuhp_step *sp;
+	int cpu;
+	int ret;
+
+	sp = cpuhp_get_step(state);
+	if (sp->multi_instance == false)
+		return -EINVAL;
+
+	get_online_cpus();
+
+	if (!invoke || !sp->startup_multi)
+		goto add_node;
+
+	/*
+	 * Try to call the startup callback for each present cpu
+	 * depending on the hotplug state of the cpu.
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate < state)
+			continue;
+
+		ret = cpuhp_issue_call(cpu, state, true, node);
+		if (ret) {
+			if (sp->teardown_multi)
+				cpuhp_rollback_install(cpu, state, node);
+			goto err;
+		}
+	}
+add_node:
+	ret = 0;
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_add_head(node, &sp->list);
+	mutex_unlock(&cpuhp_state_mutex);
+
+err:
+	put_online_cpus();
+	return ret;
+}
+
+
 /**
  * __cpuhp_setup_state - Setup the callbacks for an hotplug machine state
  * @state:	The state to setup
@@ -1482,7 +1585,8 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 int __cpuhp_setup_state(enum cpuhp_state state,
 			const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu))
+			int (*teardown)(unsigned int cpu),
+			bool multi_instance)
 {
 	int cpu, ret = 0;
 	int dyn_state = 0;
@@ -1501,7 +1605,7 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		state = ret;
 	}
 
-	cpuhp_store_callbacks(state, name, startup, teardown);
+	cpuhp_store_callbacks(state, name, startup, teardown, multi_instance);
 
 	if (!invoke || !startup)
 		goto out;
@@ -1517,11 +1621,11 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, true);
+		ret = cpuhp_issue_call(cpu, state, true, NULL);
 		if (ret) {
 			if (teardown)
-				cpuhp_rollback_install(cpu, state);
-			cpuhp_store_callbacks(state, NULL, NULL, NULL);
+				cpuhp_rollback_install(cpu, state, NULL);
+			cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 			goto out;
 		}
 	}
@@ -1533,6 +1637,42 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 }
 EXPORT_SYMBOL(__cpuhp_setup_state);
 
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke)
+{
+	struct cpuhp_step *sp = cpuhp_get_step(state);
+	int cpu;
+
+	BUG_ON(cpuhp_cb_check(state));
+
+	if (!sp->multi_instance)
+		return -EINVAL;
+
+	get_online_cpus();
+	if (!invoke || !cpuhp_get_teardown_cb(state))
+		goto remove;
+	/*
+	 * Call the teardown callback for each present cpu depending
+	 * on the hotplug state of the cpu. This function is not
+	 * allowed to fail currently!
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate >= state)
+			cpuhp_issue_call(cpu, state, false, node);
+	}
+
+remove:
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_del(node);
+	mutex_unlock(&cpuhp_state_mutex);
+	put_online_cpus();
+
+	return 0;
+}
+
 /**
  * __cpuhp_remove_state - Remove the callbacks for an hotplug machine state
  * @state:	The state to remove
@@ -1544,12 +1684,19 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
+	if (sp->multi_instance) {
+		WARN(!hlist_empty(&sp->list),
+		     "Error: Removing state %d which has instances left.\n",
+		     state);
+	}
+
 	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
@@ -1563,10 +1710,10 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, NULL);
 	}
 remove:
-	cpuhp_store_callbacks(state, NULL, NULL, NULL);
+	cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 	put_online_cpus();
 }
 EXPORT_SYMBOL(__cpuhp_remove_state);
-- 
2.8.1

[tip:smp/hotplug] cpu/hotplug: Add multi instance support

[tip-bot for Thomas Gleixner]

Commit-ID:  cf95baff0dd05eaaf445f75217a3363d4f83b449
Gitweb:     http://git.kernel.org/tip/cf95baff0dd05eaaf445f75217a3363d4f83b449
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:39 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:28 +0200

cpu/hotplug: Add multi instance support

This patch adds the ability for a given state to have multiple
instances. Until now all states have a single instance and the startup /
teardown callback use global variables.
A few drivers need to perform a the same callbacks on multiple
"instances". Currently we have three drivers in tree which all have a
global list which they iterate over. With multi instance they support
don't need their private list and the functionality has been moved into
core code. Plus we hold the hotplug lock in core so no cpus comes/goes
while instances are registered and we do rollback in error case :)

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
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/1471024183-12666-3-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 include/linux/cpuhotplug.h   |  97 ++++++++++++++++++-
 include/trace/events/cpuhp.h |  28 ++++++
 kernel/cpu.c                 | 217 ++++++++++++++++++++++++++++++++++++-------
 3 files changed, 304 insertions(+), 38 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 242bf53..81cc207 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -99,7 +99,7 @@ enum cpuhp_state {
 
 int __cpuhp_setup_state(enum cpuhp_state state,	const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu));
+			int (*teardown)(unsigned int cpu), bool multi_instance);
 
 /**
  * cpuhp_setup_state - Setup hotplug state callbacks with calling the callbacks
@@ -116,7 +116,7 @@ static inline int cpuhp_setup_state(enum cpuhp_state state,
 				    int (*startup)(unsigned int cpu),
 				    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, true, startup, teardown);
+	return __cpuhp_setup_state(state, name, true, startup, teardown, false);
 }
 
 /**
@@ -135,7 +135,66 @@ static inline int cpuhp_setup_state_nocalls(enum cpuhp_state state,
 					    int (*startup)(unsigned int cpu),
 					    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, false, startup, teardown);
+	return __cpuhp_setup_state(state, name, false, startup, teardown,
+				   false);
+}
+
+/**
+ * cpuhp_setup_state_multi - Add callbacks for multi state
+ * @state:	The state for which the calls are installed
+ * @name:	Name of the callback.
+ * @startup:	startup callback function
+ * @teardown:	teardown callback function
+ *
+ * Sets the internal multi_instance flag and prepares a state to work as a multi
+ * instance callback. No callbacks are invoked at this point. The callbacks are
+ * invoked once an instance for this state are registered via
+ * @cpuhp_state_add_instance or @cpuhp_state_add_instance_nocalls.
+ */
+static inline int cpuhp_setup_state_multi(enum cpuhp_state state,
+					  const char *name,
+					  int (*startup)(unsigned int cpu,
+							 struct hlist_node *node),
+					  int (*teardown)(unsigned int cpu,
+							  struct hlist_node *node))
+{
+	return __cpuhp_setup_state(state, name, false,
+				   (void *) startup,
+				   (void *) teardown, true);
+}
+
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke);
+
+/**
+ * cpuhp_state_add_instance - Add an instance for a state and invoke startup
+ *                            callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state and invokes the startup callback on
+ * the present cpus which have already reached the @state. The @state must have
+ * been earlier marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance(enum cpuhp_state state,
+					   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_add_instance_nocalls - Add an instance for a state without
+ *                                    invoking the startup callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state The @state must have been earlier
+ * marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance_nocalls(enum cpuhp_state state,
+						   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, false);
 }
 
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke);
@@ -162,6 +221,38 @@ static inline void cpuhp_remove_state_nocalls(enum cpuhp_state state)
 	__cpuhp_remove_state(state, false);
 }
 
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke);
+
+/**
+ * cpuhp_state_remove_instance - Remove hotplug instance from state and invoke
+ *                               the teardown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance and invokes the teardown callback on the present cpus
+ * which have already reached the @state.
+ */
+static inline int cpuhp_state_remove_instance(enum cpuhp_state state,
+					      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_remove_instance_nocalls - Remove hotplug instance from state
+ *					 without invoking the reatdown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance without invoking the teardown callback.
+ */
+static inline int cpuhp_state_remove_instance_nocalls(enum cpuhp_state state,
+						      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, false);
+}
+
 #ifdef CONFIG_SMP
 void cpuhp_online_idle(enum cpuhp_state state);
 #else
diff --git a/include/trace/events/cpuhp.h b/include/trace/events/cpuhp.h
index a72bd93..996953d 100644
--- a/include/trace/events/cpuhp.h
+++ b/include/trace/events/cpuhp.h
@@ -33,6 +33,34 @@ TRACE_EVENT(cpuhp_enter,
 		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
 );
 
+TRACE_EVENT(cpuhp_multi_enter,
+
+	TP_PROTO(unsigned int cpu,
+		 int target,
+		 int idx,
+		 int (*fun)(unsigned int, struct hlist_node *),
+		 struct hlist_node *node),
+
+	TP_ARGS(cpu, target, idx, fun, node),
+
+	TP_STRUCT__entry(
+		__field( unsigned int,	cpu		)
+		__field( int,		target		)
+		__field( int,		idx		)
+		__field( void *,	fun		)
+	),
+
+	TP_fast_assign(
+		__entry->cpu	= cpu;
+		__entry->target	= target;
+		__entry->idx	= idx;
+		__entry->fun	= fun;
+	),
+
+	TP_printk("cpu: %04u target: %3d step: %3d (%pf)",
+		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
+);
+
 TRACE_EVENT(cpuhp_exit,
 
 	TP_PROTO(unsigned int cpu,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index d36d8e0..37e8c9f 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -52,6 +52,7 @@ struct cpuhp_cpu_state {
 	bool			rollback;
 	bool			single;
 	bool			bringup;
+	struct hlist_node	*node;
 	enum cpuhp_state	cb_state;
 	int			result;
 	struct completion	done;
@@ -70,11 +71,21 @@ static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state);
  * @cant_stop:	Bringup/teardown can't be stopped at this step
  */
 struct cpuhp_step {
-	const char	*name;
-	int		(*startup)(unsigned int cpu);
-	int		(*teardown)(unsigned int cpu);
-	bool		skip_onerr;
-	bool		cant_stop;
+	const char		*name;
+	union {
+		int		(*startup)(unsigned int cpu);
+		int		(*startup_multi)(unsigned int cpu,
+						 struct hlist_node *node);
+	};
+	union {
+		int		(*teardown)(unsigned int cpu);
+		int		(*teardown_multi)(unsigned int cpu,
+						  struct hlist_node *node);
+	};
+	struct hlist_head	list;
+	bool			skip_onerr;
+	bool			cant_stop;
+	bool			multi_instance;
 };
 
 static DEFINE_MUTEX(cpuhp_state_mutex);
@@ -104,20 +115,59 @@ static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
  * @step:	The step in the state machine
  * @bringup:	True if the bringup callback should be invoked
  *
- * Called from cpu hotplug and from the state register machinery
+ * Called from cpu hotplug and from the state register machinery.
  */
 static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
-				 bool bringup)
+				 bool bringup, struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 	struct cpuhp_step *step = cpuhp_get_step(state);
-	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
-	int ret = 0;
-
-	if (cb) {
+	int (*cbm)(unsigned int cpu, struct hlist_node *node);
+	int (*cb)(unsigned int cpu);
+	int ret, cnt;
+
+	if (!step->multi_instance) {
+		cb = bringup ? step->startup : step->teardown;
+		if (!cb)
+			return 0;
 		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
 		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+	cbm = bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return 0;
+
+	/* Single invocation for instance add/remove */
+	if (node) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+
+	/* State transition. Invoke on all instances */
+	cnt = 0;
+	hlist_for_each(node, &step->list) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		if (ret)
+			goto err;
+		cnt++;
+	}
+	return 0;
+err:
+	/* Rollback the instances if one failed */
+	cbm = !bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return ret;
+
+	hlist_for_each(node, &step->list) {
+		if (!cnt--)
+			break;
+		cbm(cpu, node);
 	}
 	return ret;
 }
@@ -398,7 +448,7 @@ static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, true);
+			cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -409,7 +459,7 @@ static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		ret = cpuhp_invoke_callback(cpu, st->state, false);
+		ret = cpuhp_invoke_callback(cpu, st->state, false, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_down(cpu, st);
@@ -425,7 +475,7 @@ static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, false);
+			cpuhp_invoke_callback(cpu, st->state, false, NULL);
 	}
 }
 
@@ -437,7 +487,7 @@ static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 
 	while (st->state < target) {
 		st->state++;
-		ret = cpuhp_invoke_callback(cpu, st->state, true);
+		ret = cpuhp_invoke_callback(cpu, st->state, true, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_up(cpu, st);
@@ -502,11 +552,11 @@ static void cpuhp_thread_fun(unsigned int cpu)
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 			local_irq_enable();
 		} else {
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
@@ -534,7 +584,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 
 /* Invoke a single callback on a remote cpu */
 static int
-cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup,
+			 struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -546,11 +597,12 @@ cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, bringup);
+		return cpuhp_invoke_callback(cpu, state, bringup, node);
 
 	st->cb_state = state;
 	st->single = true;
 	st->bringup = bringup;
+	st->node = node;
 
 	/*
 	 * Make sure the above stores are visible before should_run becomes
@@ -726,7 +778,7 @@ static int take_cpu_down(void *_param)
 	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
 	for (; st->state > target; st->state--)
-		cpuhp_invoke_callback(cpu, st->state, false);
+		cpuhp_invoke_callback(cpu, st->state, false, NULL);
 
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
@@ -921,7 +973,7 @@ void notify_cpu_starting(unsigned int cpu)
 
 	while (st->state < target) {
 		st->state++;
-		cpuhp_invoke_callback(cpu, st->state, true);
+		cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -1386,7 +1438,8 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
-				  int (*teardown)(unsigned int cpu))
+				  int (*teardown)(unsigned int cpu),
+				  bool multi_instance)
 {
 	/* (Un)Install the callbacks for further cpu hotplug operations */
 	struct cpuhp_step *sp;
@@ -1396,6 +1449,8 @@ static void cpuhp_store_callbacks(enum cpuhp_state state,
 	sp->startup = startup;
 	sp->teardown = teardown;
 	sp->name = name;
+	sp->multi_instance = multi_instance;
+	INIT_HLIST_HEAD(&sp->list);
 	mutex_unlock(&cpuhp_state_mutex);
 }
 
@@ -1408,7 +1463,8 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup,
+			    struct hlist_node *node)
 {
 	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
@@ -1421,11 +1477,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup, node);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, bringup);
+	ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1436,7 +1492,8 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
+				   struct hlist_node *node)
 {
 	int cpu;
 
@@ -1450,7 +1507,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, node);
 	}
 }
 
@@ -1477,6 +1534,52 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 	return -ENOSPC;
 }
 
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke)
+{
+	struct cpuhp_step *sp;
+	int cpu;
+	int ret;
+
+	sp = cpuhp_get_step(state);
+	if (sp->multi_instance == false)
+		return -EINVAL;
+
+	get_online_cpus();
+
+	if (!invoke || !sp->startup_multi)
+		goto add_node;
+
+	/*
+	 * Try to call the startup callback for each present cpu
+	 * depending on the hotplug state of the cpu.
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate < state)
+			continue;
+
+		ret = cpuhp_issue_call(cpu, state, true, node);
+		if (ret) {
+			if (sp->teardown_multi)
+				cpuhp_rollback_install(cpu, state, node);
+			goto err;
+		}
+	}
+add_node:
+	ret = 0;
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_add_head(node, &sp->list);
+	mutex_unlock(&cpuhp_state_mutex);
+
+err:
+	put_online_cpus();
+	return ret;
+}
+
+
 /**
  * __cpuhp_setup_state - Setup the callbacks for an hotplug machine state
  * @state:	The state to setup
@@ -1490,7 +1593,8 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 int __cpuhp_setup_state(enum cpuhp_state state,
 			const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu))
+			int (*teardown)(unsigned int cpu),
+			bool multi_instance)
 {
 	int cpu, ret = 0;
 	int dyn_state = 0;
@@ -1509,7 +1613,7 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		state = ret;
 	}
 
-	cpuhp_store_callbacks(state, name, startup, teardown);
+	cpuhp_store_callbacks(state, name, startup, teardown, multi_instance);
 
 	if (!invoke || !startup)
 		goto out;
@@ -1525,11 +1629,11 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, true);
+		ret = cpuhp_issue_call(cpu, state, true, NULL);
 		if (ret) {
 			if (teardown)
-				cpuhp_rollback_install(cpu, state);
-			cpuhp_store_callbacks(state, NULL, NULL, NULL);
+				cpuhp_rollback_install(cpu, state, NULL);
+			cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 			goto out;
 		}
 	}
@@ -1541,6 +1645,42 @@ out:
 }
 EXPORT_SYMBOL(__cpuhp_setup_state);
 
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke)
+{
+	struct cpuhp_step *sp = cpuhp_get_step(state);
+	int cpu;
+
+	BUG_ON(cpuhp_cb_check(state));
+
+	if (!sp->multi_instance)
+		return -EINVAL;
+
+	get_online_cpus();
+	if (!invoke || !cpuhp_get_teardown_cb(state))
+		goto remove;
+	/*
+	 * Call the teardown callback for each present cpu depending
+	 * on the hotplug state of the cpu. This function is not
+	 * allowed to fail currently!
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate >= state)
+			cpuhp_issue_call(cpu, state, false, node);
+	}
+
+remove:
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_del(node);
+	mutex_unlock(&cpuhp_state_mutex);
+	put_online_cpus();
+
+	return 0;
+}
+
 /**
  * __cpuhp_remove_state - Remove the callbacks for an hotplug machine state
  * @state:	The state to remove
@@ -1552,12 +1692,19 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
+	if (sp->multi_instance) {
+		WARN(!hlist_empty(&sp->list),
+		     "Error: Removing state %d which has instances left.\n",
+		     state);
+	}
+
 	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
@@ -1571,10 +1718,10 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, NULL);
 	}
 remove:
-	cpuhp_store_callbacks(state, NULL, NULL, NULL);
+	cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 	put_online_cpus();
 }
 EXPORT_SYMBOL(__cpuhp_remove_state);

[tip:smp/hotplug] cpu/hotplug: Add multi instance support

[tip-bot for Thomas Gleixner]

Commit-ID:  cf392d10b69e6e6c57ceea48b347a2ab1a4b75b2
Gitweb:     http://git.kernel.org/tip/cf392d10b69e6e6c57ceea48b347a2ab1a4b75b2
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:39 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:05 +0200

cpu/hotplug: Add multi instance support

This patch adds the ability for a given state to have multiple
instances. Until now all states have a single instance and the startup /
teardown callback use global variables.
A few drivers need to perform a the same callbacks on multiple
"instances". Currently we have three drivers in tree which all have a
global list which they iterate over. With multi instance they support
don't need their private list and the functionality has been moved into
core code. Plus we hold the hotplug lock in core so no cpus comes/goes
while instances are registered and we do rollback in error case :)

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
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/1471024183-12666-3-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 include/linux/cpuhotplug.h   | 110 +++++++++++++++++++++-
 include/trace/events/cpuhp.h |  28 ++++++
 kernel/cpu.c                 | 218 ++++++++++++++++++++++++++++++++++++-------
 3 files changed, 318 insertions(+), 38 deletions(-)

diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 242bf53..dcfe6191 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -99,7 +99,7 @@ enum cpuhp_state {
 
 int __cpuhp_setup_state(enum cpuhp_state state,	const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu));
+			int (*teardown)(unsigned int cpu), bool multi_instance);
 
 /**
  * cpuhp_setup_state - Setup hotplug state callbacks with calling the callbacks
@@ -116,7 +116,7 @@ static inline int cpuhp_setup_state(enum cpuhp_state state,
 				    int (*startup)(unsigned int cpu),
 				    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, true, startup, teardown);
+	return __cpuhp_setup_state(state, name, true, startup, teardown, false);
 }
 
 /**
@@ -135,7 +135,66 @@ static inline int cpuhp_setup_state_nocalls(enum cpuhp_state state,
 					    int (*startup)(unsigned int cpu),
 					    int (*teardown)(unsigned int cpu))
 {
-	return __cpuhp_setup_state(state, name, false, startup, teardown);
+	return __cpuhp_setup_state(state, name, false, startup, teardown,
+				   false);
+}
+
+/**
+ * cpuhp_setup_state_multi - Add callbacks for multi state
+ * @state:	The state for which the calls are installed
+ * @name:	Name of the callback.
+ * @startup:	startup callback function
+ * @teardown:	teardown callback function
+ *
+ * Sets the internal multi_instance flag and prepares a state to work as a multi
+ * instance callback. No callbacks are invoked at this point. The callbacks are
+ * invoked once an instance for this state are registered via
+ * @cpuhp_state_add_instance or @cpuhp_state_add_instance_nocalls.
+ */
+static inline int cpuhp_setup_state_multi(enum cpuhp_state state,
+					  const char *name,
+					  int (*startup)(unsigned int cpu,
+							 struct hlist_node *node),
+					  int (*teardown)(unsigned int cpu,
+							  struct hlist_node *node))
+{
+	return __cpuhp_setup_state(state, name, false,
+				   (void *) startup,
+				   (void *) teardown, true);
+}
+
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke);
+
+/**
+ * cpuhp_state_add_instance - Add an instance for a state and invoke startup
+ *                            callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state and invokes the startup callback on
+ * the present cpus which have already reached the @state. The @state must have
+ * been earlier marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance(enum cpuhp_state state,
+					   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_add_instance_nocalls - Add an instance for a state without
+ *                                    invoking the startup callback.
+ * @state:	The state for which the instance is installed
+ * @node:	The node for this individual state.
+ *
+ * Installs the instance for the @state The @state must have been earlier
+ * marked as multi-instance by @cpuhp_setup_state_multi.
+ */
+static inline int cpuhp_state_add_instance_nocalls(enum cpuhp_state state,
+						   struct hlist_node *node)
+{
+	return __cpuhp_state_add_instance(state, node, false);
 }
 
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke);
@@ -162,6 +221,51 @@ static inline void cpuhp_remove_state_nocalls(enum cpuhp_state state)
 	__cpuhp_remove_state(state, false);
 }
 
+/**
+ * cpuhp_remove_multi_state - Remove hotplug multi state callback
+ * @state:	The state for which the calls are removed
+ *
+ * Removes the callback functions from a multi state. This is the reverse of
+ * cpuhp_setup_state_multi(). All instances should have been removed before
+ * invoking this function.
+ */
+static inline void cpuhp_remove_multi_state(enum cpuhp_state state)
+{
+	__cpuhp_remove_state(state, false);
+}
+
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke);
+
+/**
+ * cpuhp_state_remove_instance - Remove hotplug instance from state and invoke
+ *                               the teardown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance and invokes the teardown callback on the present cpus
+ * which have already reached the @state.
+ */
+static inline int cpuhp_state_remove_instance(enum cpuhp_state state,
+					      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, true);
+}
+
+/**
+ * cpuhp_state_remove_instance_nocalls - Remove hotplug instance from state
+ *					 without invoking the reatdown callback
+ * @state:	The state from which the instance is removed
+ * @node:	The node for this individual state.
+ *
+ * Removes the instance without invoking the teardown callback.
+ */
+static inline int cpuhp_state_remove_instance_nocalls(enum cpuhp_state state,
+						      struct hlist_node *node)
+{
+	return __cpuhp_state_remove_instance(state, node, false);
+}
+
 #ifdef CONFIG_SMP
 void cpuhp_online_idle(enum cpuhp_state state);
 #else
diff --git a/include/trace/events/cpuhp.h b/include/trace/events/cpuhp.h
index a72bd93..996953d 100644
--- a/include/trace/events/cpuhp.h
+++ b/include/trace/events/cpuhp.h
@@ -33,6 +33,34 @@ TRACE_EVENT(cpuhp_enter,
 		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
 );
 
+TRACE_EVENT(cpuhp_multi_enter,
+
+	TP_PROTO(unsigned int cpu,
+		 int target,
+		 int idx,
+		 int (*fun)(unsigned int, struct hlist_node *),
+		 struct hlist_node *node),
+
+	TP_ARGS(cpu, target, idx, fun, node),
+
+	TP_STRUCT__entry(
+		__field( unsigned int,	cpu		)
+		__field( int,		target		)
+		__field( int,		idx		)
+		__field( void *,	fun		)
+	),
+
+	TP_fast_assign(
+		__entry->cpu	= cpu;
+		__entry->target	= target;
+		__entry->idx	= idx;
+		__entry->fun	= fun;
+	),
+
+	TP_printk("cpu: %04u target: %3d step: %3d (%pf)",
+		  __entry->cpu, __entry->target, __entry->idx, __entry->fun)
+);
+
 TRACE_EVENT(cpuhp_exit,
 
 	TP_PROTO(unsigned int cpu,
diff --git a/kernel/cpu.c b/kernel/cpu.c
index d36d8e0..c506485 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -52,6 +52,7 @@ struct cpuhp_cpu_state {
 	bool			rollback;
 	bool			single;
 	bool			bringup;
+	struct hlist_node	*node;
 	enum cpuhp_state	cb_state;
 	int			result;
 	struct completion	done;
@@ -70,11 +71,21 @@ static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state);
  * @cant_stop:	Bringup/teardown can't be stopped at this step
  */
 struct cpuhp_step {
-	const char	*name;
-	int		(*startup)(unsigned int cpu);
-	int		(*teardown)(unsigned int cpu);
-	bool		skip_onerr;
-	bool		cant_stop;
+	const char		*name;
+	union {
+		int		(*startup)(unsigned int cpu);
+		int		(*startup_multi)(unsigned int cpu,
+						 struct hlist_node *node);
+	};
+	union {
+		int		(*teardown)(unsigned int cpu);
+		int		(*teardown_multi)(unsigned int cpu,
+						  struct hlist_node *node);
+	};
+	struct hlist_head	list;
+	bool			skip_onerr;
+	bool			cant_stop;
+	bool			multi_instance;
 };
 
 static DEFINE_MUTEX(cpuhp_state_mutex);
@@ -104,20 +115,59 @@ static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
  * @step:	The step in the state machine
  * @bringup:	True if the bringup callback should be invoked
  *
- * Called from cpu hotplug and from the state register machinery
+ * Called from cpu hotplug and from the state register machinery.
  */
 static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
-				 bool bringup)
+				 bool bringup, struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 	struct cpuhp_step *step = cpuhp_get_step(state);
-	int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown;
-	int ret = 0;
-
-	if (cb) {
+	int (*cbm)(unsigned int cpu, struct hlist_node *node);
+	int (*cb)(unsigned int cpu);
+	int ret, cnt;
+
+	if (!step->multi_instance) {
+		cb = bringup ? step->startup : step->teardown;
+		if (!cb)
+			return 0;
 		trace_cpuhp_enter(cpu, st->target, state, cb);
 		ret = cb(cpu);
 		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+	cbm = bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return 0;
+
+	/* Single invocation for instance add/remove */
+	if (node) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		return ret;
+	}
+
+	/* State transition. Invoke on all instances */
+	cnt = 0;
+	hlist_for_each(node, &step->list) {
+		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
+		ret = cbm(cpu, node);
+		trace_cpuhp_exit(cpu, st->state, state, ret);
+		if (ret)
+			goto err;
+		cnt++;
+	}
+	return 0;
+err:
+	/* Rollback the instances if one failed */
+	cbm = !bringup ? step->startup_multi : step->teardown_multi;
+	if (!cbm)
+		return ret;
+
+	hlist_for_each(node, &step->list) {
+		if (!cnt--)
+			break;
+		cbm(cpu, node);
 	}
 	return ret;
 }
@@ -398,7 +448,7 @@ static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, true);
+			cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -409,7 +459,7 @@ static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 	int ret = 0;
 
 	for (; st->state > target; st->state--) {
-		ret = cpuhp_invoke_callback(cpu, st->state, false);
+		ret = cpuhp_invoke_callback(cpu, st->state, false, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_down(cpu, st);
@@ -425,7 +475,7 @@ static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
 		struct cpuhp_step *step = cpuhp_get_step(st->state);
 
 		if (!step->skip_onerr)
-			cpuhp_invoke_callback(cpu, st->state, false);
+			cpuhp_invoke_callback(cpu, st->state, false, NULL);
 	}
 }
 
@@ -437,7 +487,7 @@ static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
 
 	while (st->state < target) {
 		st->state++;
-		ret = cpuhp_invoke_callback(cpu, st->state, true);
+		ret = cpuhp_invoke_callback(cpu, st->state, true, NULL);
 		if (ret) {
 			st->target = prev_state;
 			undo_cpu_up(cpu, st);
@@ -502,11 +552,11 @@ static void cpuhp_thread_fun(unsigned int cpu)
 		if (st->cb_state < CPUHP_AP_ONLINE) {
 			local_irq_disable();
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 			local_irq_enable();
 		} else {
 			ret = cpuhp_invoke_callback(cpu, st->cb_state,
-						    st->bringup);
+						    st->bringup, st->node);
 		}
 	} else if (st->rollback) {
 		BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE);
@@ -534,7 +584,8 @@ static void cpuhp_thread_fun(unsigned int cpu)
 
 /* Invoke a single callback on a remote cpu */
 static int
-cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
+cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup,
+			 struct hlist_node *node)
 {
 	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
 
@@ -546,11 +597,12 @@ cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup)
 	 * we invoke the thread function directly.
 	 */
 	if (!st->thread)
-		return cpuhp_invoke_callback(cpu, state, bringup);
+		return cpuhp_invoke_callback(cpu, state, bringup, node);
 
 	st->cb_state = state;
 	st->single = true;
 	st->bringup = bringup;
+	st->node = node;
 
 	/*
 	 * Make sure the above stores are visible before should_run becomes
@@ -726,7 +778,7 @@ static int take_cpu_down(void *_param)
 	st->state--;
 	/* Invoke the former CPU_DYING callbacks */
 	for (; st->state > target; st->state--)
-		cpuhp_invoke_callback(cpu, st->state, false);
+		cpuhp_invoke_callback(cpu, st->state, false, NULL);
 
 	/* Give up timekeeping duties */
 	tick_handover_do_timer();
@@ -921,7 +973,7 @@ void notify_cpu_starting(unsigned int cpu)
 
 	while (st->state < target) {
 		st->state++;
-		cpuhp_invoke_callback(cpu, st->state, true);
+		cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
@@ -1386,7 +1438,8 @@ static int cpuhp_cb_check(enum cpuhp_state state)
 static void cpuhp_store_callbacks(enum cpuhp_state state,
 				  const char *name,
 				  int (*startup)(unsigned int cpu),
-				  int (*teardown)(unsigned int cpu))
+				  int (*teardown)(unsigned int cpu),
+				  bool multi_instance)
 {
 	/* (Un)Install the callbacks for further cpu hotplug operations */
 	struct cpuhp_step *sp;
@@ -1396,6 +1449,8 @@ static void cpuhp_store_callbacks(enum cpuhp_state state,
 	sp->startup = startup;
 	sp->teardown = teardown;
 	sp->name = name;
+	sp->multi_instance = multi_instance;
+	INIT_HLIST_HEAD(&sp->list);
 	mutex_unlock(&cpuhp_state_mutex);
 }
 
@@ -1408,7 +1463,8 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
  * Call the startup/teardown function for a step either on the AP or
  * on the current CPU.
  */
-static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
+static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup,
+			    struct hlist_node *node)
 {
 	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int ret;
@@ -1421,11 +1477,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
 	 */
 #ifdef CONFIG_SMP
 	if (cpuhp_is_ap_state(state))
-		ret = cpuhp_invoke_ap_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_ap_callback(cpu, state, bringup, node);
 	else
-		ret = cpuhp_invoke_callback(cpu, state, bringup);
+		ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #else
-	ret = cpuhp_invoke_callback(cpu, state, bringup);
+	ret = cpuhp_invoke_callback(cpu, state, bringup, node);
 #endif
 	BUG_ON(ret && !bringup);
 	return ret;
@@ -1436,7 +1492,8 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup)
  *
  * Note: The teardown callbacks for rollback are not allowed to fail!
  */
-static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
+static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
+				   struct hlist_node *node)
 {
 	int cpu;
 
@@ -1450,7 +1507,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state)
 
 		/* Did we invoke the startup call on that cpu ? */
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, node);
 	}
 }
 
@@ -1477,6 +1534,52 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 	return -ENOSPC;
 }
 
+int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
+			       bool invoke)
+{
+	struct cpuhp_step *sp;
+	int cpu;
+	int ret;
+
+	sp = cpuhp_get_step(state);
+	if (sp->multi_instance == false)
+		return -EINVAL;
+
+	get_online_cpus();
+
+	if (!invoke || !sp->startup_multi)
+		goto add_node;
+
+	/*
+	 * Try to call the startup callback for each present cpu
+	 * depending on the hotplug state of the cpu.
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate < state)
+			continue;
+
+		ret = cpuhp_issue_call(cpu, state, true, node);
+		if (ret) {
+			if (sp->teardown_multi)
+				cpuhp_rollback_install(cpu, state, node);
+			goto err;
+		}
+	}
+add_node:
+	ret = 0;
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_add_head(node, &sp->list);
+	mutex_unlock(&cpuhp_state_mutex);
+
+err:
+	put_online_cpus();
+	return ret;
+}
+EXPORT_SYMBOL_GPL(__cpuhp_state_add_instance);
+
 /**
  * __cpuhp_setup_state - Setup the callbacks for an hotplug machine state
  * @state:	The state to setup
@@ -1490,7 +1593,8 @@ static int cpuhp_reserve_state(enum cpuhp_state state)
 int __cpuhp_setup_state(enum cpuhp_state state,
 			const char *name, bool invoke,
 			int (*startup)(unsigned int cpu),
-			int (*teardown)(unsigned int cpu))
+			int (*teardown)(unsigned int cpu),
+			bool multi_instance)
 {
 	int cpu, ret = 0;
 	int dyn_state = 0;
@@ -1509,7 +1613,7 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		state = ret;
 	}
 
-	cpuhp_store_callbacks(state, name, startup, teardown);
+	cpuhp_store_callbacks(state, name, startup, teardown, multi_instance);
 
 	if (!invoke || !startup)
 		goto out;
@@ -1525,11 +1629,11 @@ int __cpuhp_setup_state(enum cpuhp_state state,
 		if (cpustate < state)
 			continue;
 
-		ret = cpuhp_issue_call(cpu, state, true);
+		ret = cpuhp_issue_call(cpu, state, true, NULL);
 		if (ret) {
 			if (teardown)
-				cpuhp_rollback_install(cpu, state);
-			cpuhp_store_callbacks(state, NULL, NULL, NULL);
+				cpuhp_rollback_install(cpu, state, NULL);
+			cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 			goto out;
 		}
 	}
@@ -1541,6 +1645,42 @@ out:
 }
 EXPORT_SYMBOL(__cpuhp_setup_state);
 
+int __cpuhp_state_remove_instance(enum cpuhp_state state,
+				  struct hlist_node *node, bool invoke)
+{
+	struct cpuhp_step *sp = cpuhp_get_step(state);
+	int cpu;
+
+	BUG_ON(cpuhp_cb_check(state));
+
+	if (!sp->multi_instance)
+		return -EINVAL;
+
+	get_online_cpus();
+	if (!invoke || !cpuhp_get_teardown_cb(state))
+		goto remove;
+	/*
+	 * Call the teardown callback for each present cpu depending
+	 * on the hotplug state of the cpu. This function is not
+	 * allowed to fail currently!
+	 */
+	for_each_present_cpu(cpu) {
+		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
+		int cpustate = st->state;
+
+		if (cpustate >= state)
+			cpuhp_issue_call(cpu, state, false, node);
+	}
+
+remove:
+	mutex_lock(&cpuhp_state_mutex);
+	hlist_del(node);
+	mutex_unlock(&cpuhp_state_mutex);
+	put_online_cpus();
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(__cpuhp_state_remove_instance);
 /**
  * __cpuhp_remove_state - Remove the callbacks for an hotplug machine state
  * @state:	The state to remove
@@ -1552,12 +1692,20 @@ EXPORT_SYMBOL(__cpuhp_setup_state);
  */
 void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 {
+	struct cpuhp_step *sp = cpuhp_get_step(state);
 	int cpu;
 
 	BUG_ON(cpuhp_cb_check(state));
 
 	get_online_cpus();
 
+	if (sp->multi_instance) {
+		WARN(!hlist_empty(&sp->list),
+		     "Error: Removing state %d which has instances left.\n",
+		     state);
+		goto remove;
+	}
+
 	if (!invoke || !cpuhp_get_teardown_cb(state))
 		goto remove;
 
@@ -1571,10 +1719,10 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
 		int cpustate = st->state;
 
 		if (cpustate >= state)
-			cpuhp_issue_call(cpu, state, false);
+			cpuhp_issue_call(cpu, state, false, NULL);
 	}
 remove:
-	cpuhp_store_callbacks(state, NULL, NULL, NULL);
+	cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
 	put_online_cpus();
 }
 EXPORT_SYMBOL(__cpuhp_remove_state);

[PATCH 3/6] arm/perf: Use multi instance instead of custom list

[Sebastian Andrzej Siewior]

Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/perf/arm_pmu.c       | 44 ++++++++++++++++++--------------------------
 include/linux/perf/arm_pmu.h |  2 +-
 2 files changed, 19 insertions(+), 27 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index 6ccb994bdfcb..9f9f4b5debdc 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -688,28 +688,20 @@ 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 arm_perf_starting_cpu(unsigned int cpu)
+static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_pmu *pmu;
+	struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
 
-	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);
+	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+		return 0;
+	if (pmu->reset)
+		pmu->reset(pmu);
 	return 0;
 }
 
@@ -821,9 +813,10 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	mutex_lock(&arm_pmu_mutex);
-	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
-	mutex_unlock(&arm_pmu_mutex);
+	err = cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					       &cpu_pmu->node);
+	if (err)
+		goto out_free;
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -859,9 +852,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	mutex_lock(&arm_pmu_mutex);
-	list_del(&cpu_pmu->entry);
-	mutex_unlock(&arm_pmu_mutex);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
+out_free:
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -869,9 +862,8 @@ 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);
-	mutex_lock(&arm_pmu_mutex);
-	list_del(&cpu_pmu->entry);
-	mutex_unlock(&arm_pmu_mutex);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1067,9 +1059,9 @@ 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);
+	ret = cpuhp_setup_state_multi(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);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index e18843809eec..4ad1b408c0bb 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 list_head	entry;
+	struct hlist_node	node;
 	struct notifier_block	cpu_pm_nb;
 };
 
-- 
2.8.1

[PATCH v2 3/6] arm/perf: Use multi instance instead of custom list

[Sebastian Andrzej Siewior]

Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
v1…v2: refresh to linux-next

 drivers/perf/arm_pmu.c       | 44 ++++++++++++++++++--------------------------
 include/linux/perf/arm_pmu.h |  2 +-
 2 files changed, 19 insertions(+), 27 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index c494613c1909..b2f742f84111 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -688,28 +688,20 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
-static DEFINE_SPINLOCK(arm_pmu_lock);
-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 arm_perf_starting_cpu(unsigned int cpu)
+static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_pmu *pmu;
+	struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
 
-	spin_lock(&arm_pmu_lock);
-	list_for_each_entry(pmu, &arm_pmu_list, entry) {
-
-		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-			continue;
-		if (pmu->reset)
-			pmu->reset(pmu);
-	}
-	spin_unlock(&arm_pmu_lock);
+	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+		return 0;
+	if (pmu->reset)
+		pmu->reset(pmu);
 	return 0;
 }
 
@@ -821,9 +813,10 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	spin_lock(&arm_pmu_lock);
-	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
-	spin_unlock(&arm_pmu_lock);
+	err = cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					       &cpu_pmu->node);
+	if (err)
+		goto out_free;
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -859,9 +852,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
+out_free:
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -869,9 +862,8 @@ 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);
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1068,9 +1060,9 @@ 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);
+	ret = cpuhp_setup_state_multi(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);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index e18843809eec..4ad1b408c0bb 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 list_head	entry;
+	struct hlist_node	node;
 	struct notifier_block	cpu_pm_nb;
 };
 
-- 
2.9.3

Re: [PATCH v2 3/6] arm/perf: Use multi instance instead of custom list

[Will Deacon]

On Wed, Aug 17, 2016 at 07:14:20PM +0200, Sebastian Andrzej Siewior wrote:
> Cc: Will Deacon <will.deacon@arm.com>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> ---
> v1…v2: refresh to linux-next
> 
>  drivers/perf/arm_pmu.c       | 44 ++++++++++++++++++--------------------------
>  include/linux/perf/arm_pmu.h |  2 +-
>  2 files changed, 19 insertions(+), 27 deletions(-)

Looks good to me:

Acked-by: Will Deacon <will.deacon@arm.com>

I'm assuming you'll take this with the rest of the series, once the
core parts have been acked.

Will

Re: [PATCH v2 3/6] arm/perf: Use multi instance instead of custom list

[Sebastian Andrzej Siewior]

On 2016-08-22 16:07:18 [+0100], Will Deacon wrote:
> Looks good to me:
> 
> Acked-by: Will Deacon <will.deacon@arm.com>
Thanks.

> I'm assuming you'll take this with the rest of the series, once the
> core parts have been acked.

I hope so, yes.

> Will

Sebastian

[tip:smp/hotplug] arm/perf: Use multi instance instead of custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  cc727977acb0fe05b7aa1f00cccb893530820895
Gitweb:     http://git.kernel.org/tip/cc727977acb0fe05b7aa1f00cccb893530820895
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 17 Aug 2016 19:14:20 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:28 +0200

arm/perf: Use multi instance instead of custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160817171420.sdwk2qivxunzryz4@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/perf/arm_pmu.c       | 44 ++++++++++++++++++--------------------------
 include/linux/perf/arm_pmu.h |  2 +-
 2 files changed, 19 insertions(+), 27 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index c494613..b2f742f 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -688,28 +688,20 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
-static DEFINE_SPINLOCK(arm_pmu_lock);
-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 arm_perf_starting_cpu(unsigned int cpu)
+static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_pmu *pmu;
-
-	spin_lock(&arm_pmu_lock);
-	list_for_each_entry(pmu, &arm_pmu_list, entry) {
+	struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
 
-		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-			continue;
-		if (pmu->reset)
-			pmu->reset(pmu);
-	}
-	spin_unlock(&arm_pmu_lock);
+	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+		return 0;
+	if (pmu->reset)
+		pmu->reset(pmu);
 	return 0;
 }
 
@@ -821,9 +813,10 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	spin_lock(&arm_pmu_lock);
-	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
-	spin_unlock(&arm_pmu_lock);
+	err = cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					       &cpu_pmu->node);
+	if (err)
+		goto out_free;
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -859,9 +852,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
+out_free:
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -869,9 +862,8 @@ out_unregister:
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1068,9 +1060,9 @@ 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);
+	ret = cpuhp_setup_state_multi(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);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index e188438..4ad1b40 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 list_head	entry;
+	struct hlist_node	node;
 	struct notifier_block	cpu_pm_nb;
 };
 

[tip:smp/hotplug] arm/perf: Use multi instance instead of custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  6e103c0cfeb9ab8d40822a015da9769595096411
Gitweb:     http://git.kernel.org/tip/6e103c0cfeb9ab8d40822a015da9769595096411
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Wed, 17 Aug 2016 19:14:20 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:06 +0200

arm/perf: Use multi instance instead of custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160817171420.sdwk2qivxunzryz4@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/perf/arm_pmu.c       | 44 ++++++++++++++++++--------------------------
 include/linux/perf/arm_pmu.h |  2 +-
 2 files changed, 19 insertions(+), 27 deletions(-)

diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index c494613..b2f742f 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -688,28 +688,20 @@ static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler)
 	return 0;
 }
 
-static DEFINE_SPINLOCK(arm_pmu_lock);
-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 arm_perf_starting_cpu(unsigned int cpu)
+static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_pmu *pmu;
-
-	spin_lock(&arm_pmu_lock);
-	list_for_each_entry(pmu, &arm_pmu_list, entry) {
+	struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
 
-		if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
-			continue;
-		if (pmu->reset)
-			pmu->reset(pmu);
-	}
-	spin_unlock(&arm_pmu_lock);
+	if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
+		return 0;
+	if (pmu->reset)
+		pmu->reset(pmu);
 	return 0;
 }
 
@@ -821,9 +813,10 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	if (!cpu_hw_events)
 		return -ENOMEM;
 
-	spin_lock(&arm_pmu_lock);
-	list_add_tail(&cpu_pmu->entry, &arm_pmu_list);
-	spin_unlock(&arm_pmu_lock);
+	err = cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					       &cpu_pmu->node);
+	if (err)
+		goto out_free;
 
 	err = cpu_pm_pmu_register(cpu_pmu);
 	if (err)
@@ -859,9 +852,9 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
 	return 0;
 
 out_unregister:
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
+out_free:
 	free_percpu(cpu_hw_events);
 	return err;
 }
@@ -869,9 +862,8 @@ out_unregister:
 static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
 {
 	cpu_pm_pmu_unregister(cpu_pmu);
-	spin_lock(&arm_pmu_lock);
-	list_del(&cpu_pmu->entry);
-	spin_unlock(&arm_pmu_lock);
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
+					    &cpu_pmu->node);
 	free_percpu(cpu_pmu->hw_events);
 }
 
@@ -1068,9 +1060,9 @@ 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);
+	ret = cpuhp_setup_state_multi(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);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index e188438..4ad1b40 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 list_head	entry;
+	struct hlist_node	node;
 	struct notifier_block	cpu_pm_nb;
 };
 

[PATCH 4/6] bus/arm-cci: Use cpu-hp's multi instance support instead custom list

[Sebastian Andrzej Siewior]

Cc: Will Deacon <will.deacon@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Olof Johansson <olof@lixom.net>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/bus/arm-cci.c | 45 ++++++++++++++++++---------------------------
 1 file changed, 18 insertions(+), 27 deletions(-)

diff --git a/drivers/bus/arm-cci.c b/drivers/bus/arm-cci.c
index 5755907f836f..4c44ba2d3412 100644
--- a/drivers/bus/arm-cci.c
+++ b/drivers/bus/arm-cci.c
@@ -144,15 +144,12 @@ struct cci_pmu {
 	int num_cntrs;
 	atomic_t active_events;
 	struct mutex reserve_mutex;
-	struct list_head entry;
+	struct hlist_node node;
 	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,
@@ -1506,25 +1503,21 @@ 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_offline_cpu(unsigned int cpu)
+static int cci_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct cci_pmu *cci_pmu;
+	struct cci_pmu *cci_pmu = hlist_entry_safe(node, struct cci_pmu, node);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		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);
-	}
-	mutex_unlock(&cci_pmu_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &cci_pmu->cpus))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	/*
+	 * TODO: migrate context once core races on event->ctx have
+	 * been fixed.
+	 */
+	cpumask_set_cpu(target, &cci_pmu->cpus);
 	return 0;
 }
 
@@ -1768,10 +1761,8 @@ static int cci_pmu_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	mutex_lock(&cci_pmu_mutex);
-	list_add(&cci_pmu->entry, &cci_pmu_list);
-	mutex_unlock(&cci_pmu_mutex);
-
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+					 &cci_pmu->node);
 	pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
 	return 0;
 }
@@ -1804,9 +1795,9 @@ 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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+				      "AP_PERF_ARM_CCI_ONLINE", NULL,
+				      cci_pmu_offline_cpu);
 	if (ret)
 		return ret;
 
-- 
2.8.1

[tip:smp/hotplug] bus/arm-cci: Use cpu-hp's multi instance support instead custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  29d6e64595f7fddb3c418a190f71409b0ff6bae8
Gitweb:     http://git.kernel.org/tip/29d6e64595f7fddb3c418a190f71409b0ff6bae8
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:41 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:29 +0200

bus/arm-cci: Use cpu-hp's multi instance support instead custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Cc: Olof Johansson <olof@lixom.net>
Link: http://lkml.kernel.org/r/1471024183-12666-5-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/bus/arm-cci.c | 45 ++++++++++++++++++---------------------------
 1 file changed, 18 insertions(+), 27 deletions(-)

diff --git a/drivers/bus/arm-cci.c b/drivers/bus/arm-cci.c
index 5755907f..4c44ba2 100644
--- a/drivers/bus/arm-cci.c
+++ b/drivers/bus/arm-cci.c
@@ -144,15 +144,12 @@ struct cci_pmu {
 	int num_cntrs;
 	atomic_t active_events;
 	struct mutex reserve_mutex;
-	struct list_head entry;
+	struct hlist_node node;
 	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,
@@ -1506,25 +1503,21 @@ 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_offline_cpu(unsigned int cpu)
+static int cci_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct cci_pmu *cci_pmu;
+	struct cci_pmu *cci_pmu = hlist_entry_safe(node, struct cci_pmu, node);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		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);
-	}
-	mutex_unlock(&cci_pmu_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &cci_pmu->cpus))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	/*
+	 * TODO: migrate context once core races on event->ctx have
+	 * been fixed.
+	 */
+	cpumask_set_cpu(target, &cci_pmu->cpus);
 	return 0;
 }
 
@@ -1768,10 +1761,8 @@ static int cci_pmu_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	mutex_lock(&cci_pmu_mutex);
-	list_add(&cci_pmu->entry, &cci_pmu_list);
-	mutex_unlock(&cci_pmu_mutex);
-
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+					 &cci_pmu->node);
 	pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
 	return 0;
 }
@@ -1804,9 +1795,9 @@ 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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+				      "AP_PERF_ARM_CCI_ONLINE", NULL,
+				      cci_pmu_offline_cpu);
 	if (ret)
 		return ret;
 

[tip:smp/hotplug] bus/arm-cci: Use cpu-hp's multi instance support instead custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  b230f0db913136f465a951806f2978b179df95d5
Gitweb:     http://git.kernel.org/tip/b230f0db913136f465a951806f2978b179df95d5
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:41 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:06 +0200

bus/arm-cci: Use cpu-hp's multi instance support instead custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: rt@linutronix.de
Cc: Olof Johansson <olof@lixom.net>
Link: http://lkml.kernel.org/r/1471024183-12666-5-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/bus/arm-cci.c | 45 ++++++++++++++++++---------------------------
 1 file changed, 18 insertions(+), 27 deletions(-)

diff --git a/drivers/bus/arm-cci.c b/drivers/bus/arm-cci.c
index 5755907f..4c44ba2 100644
--- a/drivers/bus/arm-cci.c
+++ b/drivers/bus/arm-cci.c
@@ -144,15 +144,12 @@ struct cci_pmu {
 	int num_cntrs;
 	atomic_t active_events;
 	struct mutex reserve_mutex;
-	struct list_head entry;
+	struct hlist_node node;
 	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,
@@ -1506,25 +1503,21 @@ 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_offline_cpu(unsigned int cpu)
+static int cci_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct cci_pmu *cci_pmu;
+	struct cci_pmu *cci_pmu = hlist_entry_safe(node, struct cci_pmu, node);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		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);
-	}
-	mutex_unlock(&cci_pmu_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &cci_pmu->cpus))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	/*
+	 * TODO: migrate context once core races on event->ctx have
+	 * been fixed.
+	 */
+	cpumask_set_cpu(target, &cci_pmu->cpus);
 	return 0;
 }
 
@@ -1768,10 +1761,8 @@ static int cci_pmu_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	mutex_lock(&cci_pmu_mutex);
-	list_add(&cci_pmu->entry, &cci_pmu_list);
-	mutex_unlock(&cci_pmu_mutex);
-
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+					 &cci_pmu->node);
 	pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
 	return 0;
 }
@@ -1804,9 +1795,9 @@ 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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCI_ONLINE,
+				      "AP_PERF_ARM_CCI_ONLINE", NULL,
+				      cci_pmu_offline_cpu);
 	if (ret)
 		return ret;
 

[PATCH 5/6] bus/arm-ccn: Use cpu-hp's multi instance support instead custom list

[Sebastian Andrzej Siewior]

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/bus/arm-ccn.c | 52 ++++++++++++++++++++-------------------------------
 1 file changed, 20 insertions(+), 32 deletions(-)

diff --git a/drivers/bus/arm-ccn.c b/drivers/bus/arm-ccn.c
index 97a9185af433..fcf29f416286 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 list_head entry;
+	struct hlist_node node;
 
 	struct pmu pmu;
 };
@@ -189,9 +189,6 @@ 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)
 {
 	return node / CCN_NUM_XP_PORTS;
@@ -1173,30 +1170,24 @@ static enum hrtimer_restart arm_ccn_pmu_timer_handler(struct hrtimer *hrtimer)
 }
 
 
-static int arm_ccn_pmu_offline_cpu(unsigned int cpu)
+static int arm_ccn_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_ccn_dt *dt;
+	struct arm_ccn_dt *dt = hlist_entry_safe(node, struct arm_ccn_dt, node);
+	struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		if (target >= nr_cpu_ids)
-			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);
-	}
-	mutex_unlock(&arm_ccn_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &dt->cpu))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	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);
 	return 0;
 }
 
-
 static DEFINE_IDA(arm_ccn_pmu_ida);
 
 static int arm_ccn_pmu_init(struct arm_ccn *ccn)
@@ -1278,9 +1269,8 @@ 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);
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					 &ccn->dt.node);
 	return 0;
 
 error_pmu_register:
@@ -1296,10 +1286,8 @@ 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);
-
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					    &ccn->dt.node);
 	if (ccn->irq)
 		irq_set_affinity_hint(ccn->irq, NULL);
 	for (i = 0; i < ccn->num_xps; i++)
@@ -1527,9 +1515,9 @@ static int __init arm_ccn_init(void)
 {
 	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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+				      "AP_PERF_ARM_CCN_ONLINE", NULL,
+				      arm_ccn_pmu_offline_cpu);
 	if (ret)
 		return ret;
 
-- 
2.8.1

[tip:smp/hotplug] bus/arm-ccn: Use cpu-hp's multi instance support instead custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  ebade02256fd8a449445c7b27eacb093e22d7ee5
Gitweb:     http://git.kernel.org/tip/ebade02256fd8a449445c7b27eacb093e22d7ee5
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:42 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:29 +0200

bus/arm-ccn: Use cpu-hp's multi instance support instead custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Suzuki K Poulose <suzuki.poulose@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/1471024183-12666-6-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/bus/arm-ccn.c | 52 ++++++++++++++++++++-------------------------------
 1 file changed, 20 insertions(+), 32 deletions(-)

diff --git a/drivers/bus/arm-ccn.c b/drivers/bus/arm-ccn.c
index 97a9185..fcf29f4 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 list_head entry;
+	struct hlist_node node;
 
 	struct pmu pmu;
 };
@@ -189,9 +189,6 @@ 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)
 {
 	return node / CCN_NUM_XP_PORTS;
@@ -1173,30 +1170,24 @@ static enum hrtimer_restart arm_ccn_pmu_timer_handler(struct hrtimer *hrtimer)
 }
 
 
-static int arm_ccn_pmu_offline_cpu(unsigned int cpu)
+static int arm_ccn_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_ccn_dt *dt;
+	struct arm_ccn_dt *dt = hlist_entry_safe(node, struct arm_ccn_dt, node);
+	struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		if (target >= nr_cpu_ids)
-			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);
-	}
-	mutex_unlock(&arm_ccn_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &dt->cpu))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	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);
 	return 0;
 }
 
-
 static DEFINE_IDA(arm_ccn_pmu_ida);
 
 static int arm_ccn_pmu_init(struct arm_ccn *ccn)
@@ -1278,9 +1269,8 @@ 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);
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					 &ccn->dt.node);
 	return 0;
 
 error_pmu_register:
@@ -1296,10 +1286,8 @@ 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);
-
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					    &ccn->dt.node);
 	if (ccn->irq)
 		irq_set_affinity_hint(ccn->irq, NULL);
 	for (i = 0; i < ccn->num_xps; i++)
@@ -1527,9 +1515,9 @@ static int __init arm_ccn_init(void)
 {
 	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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+				      "AP_PERF_ARM_CCN_ONLINE", NULL,
+				      arm_ccn_pmu_offline_cpu);
 	if (ret)
 		return ret;
 

[tip:smp/hotplug] bus/arm-ccn: Use cpu-hp's multi instance support instead custom list

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  8df038725ad5351a9730759e0a24a5c5d96be661
Gitweb:     http://git.kernel.org/tip/8df038725ad5351a9730759e0a24a5c5d96be661
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:42 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:06 +0200

bus/arm-ccn: Use cpu-hp's multi instance support instead custom list

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Suzuki K Poulose <suzuki.poulose@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/1471024183-12666-6-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/bus/arm-ccn.c | 54 ++++++++++++++++++++-------------------------------
 1 file changed, 21 insertions(+), 33 deletions(-)

diff --git a/drivers/bus/arm-ccn.c b/drivers/bus/arm-ccn.c
index 97a9185..e0ad475 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 list_head entry;
+	struct hlist_node node;
 
 	struct pmu pmu;
 };
@@ -189,9 +189,6 @@ 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)
 {
 	return node / CCN_NUM_XP_PORTS;
@@ -1173,30 +1170,24 @@ static enum hrtimer_restart arm_ccn_pmu_timer_handler(struct hrtimer *hrtimer)
 }
 
 
-static int arm_ccn_pmu_offline_cpu(unsigned int cpu)
+static int arm_ccn_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
-	struct arm_ccn_dt *dt;
+	struct arm_ccn_dt *dt = hlist_entry_safe(node, struct arm_ccn_dt, node);
+	struct arm_ccn *ccn = container_of(dt, struct arm_ccn, dt);
 	unsigned int target;
 
-	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))
-			continue;
-		target = cpumask_any_but(cpu_online_mask, cpu);
-		if (target >= nr_cpu_ids)
-			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);
-	}
-	mutex_unlock(&arm_ccn_mutex);
+	if (!cpumask_test_and_clear_cpu(cpu, &dt->cpu))
+		return 0;
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+	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);
 	return 0;
 }
 
-
 static DEFINE_IDA(arm_ccn_pmu_ida);
 
 static int arm_ccn_pmu_init(struct arm_ccn *ccn)
@@ -1278,9 +1269,8 @@ 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);
+	cpuhp_state_add_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					 &ccn->dt.node);
 	return 0;
 
 error_pmu_register:
@@ -1296,10 +1286,8 @@ 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);
-
+	cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+					    &ccn->dt.node);
 	if (ccn->irq)
 		irq_set_affinity_hint(ccn->irq, NULL);
 	for (i = 0; i < ccn->num_xps; i++)
@@ -1527,9 +1515,9 @@ static int __init arm_ccn_init(void)
 {
 	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);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_CCN_ONLINE,
+				      "AP_PERF_ARM_CCN_ONLINE", NULL,
+				      arm_ccn_pmu_offline_cpu);
 	if (ret)
 		return ret;
 
@@ -1541,7 +1529,7 @@ static int __init arm_ccn_init(void)
 
 static void __exit arm_ccn_exit(void)
 {
-	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_ARM_CCN_ONLINE);
+	cpuhp_remove_multi_state(CPUHP_AP_PERF_ARM_CCN_ONLINE);
 	platform_driver_unregister(&arm_ccn_driver);
 }
 

[PATCH 6/6] net: virtio-net: Convert to hotplug state machine

[Sebastian Andrzej Siewior]

Install the callbacks via the state machine.
The driver supports multiple instances and therefore the new
cpuhp_state_add_instance_nocalls() infrastrucure is used. The driver
currently uses get_online_cpus() to avoid missing a CPU hotplug event
while invoking virtnet_set_affinity(). This could be avoided by using
cpuhp_state_add_instance() variant which holds the hotplug lock and
invokes callback during registration. This is more or less a 1:1
conversation of the current code.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: virtualization@lists.linux-foundation.org
Cc: netdev@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/net/virtio_net.c   | 110 +++++++++++++++++++++++++++++++++++----------
 include/linux/cpuhotplug.h |   1 +
 2 files changed, 87 insertions(+), 24 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 1b5f531eeb25..e9be88cd76c1 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -138,8 +138,9 @@ struct virtnet_info {
 	/* Does the affinity hint is set for virtqueues? */
 	bool affinity_hint_set;
 
-	/* CPU hot plug notifier */
-	struct notifier_block nb;
+	/* CPU hotplug instances for online & dead */
+	struct hlist_node node;
+	struct hlist_node node_dead;
 
 	/* Control VQ buffers: protected by the rtnl lock */
 	struct virtio_net_ctrl_hdr ctrl_hdr;
@@ -1237,25 +1238,53 @@ static void virtnet_set_affinity(struct virtnet_info *vi)
 	vi->affinity_hint_set = true;
 }
 
-static int virtnet_cpu_callback(struct notifier_block *nfb,
-			        unsigned long action, void *hcpu)
+static int virtnet_cpu_online(unsigned int cpu, struct hlist_node *node)
 {
-	struct virtnet_info *vi = container_of(nfb, struct virtnet_info, nb);
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	switch(action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-	case CPU_DEAD:
-		virtnet_set_affinity(vi);
-		break;
-	case CPU_DOWN_PREPARE:
-		virtnet_clean_affinity(vi, (long)hcpu);
-		break;
-	default:
-		break;
-	}
+static int virtnet_cpu_dead(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node_dead);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	return NOTIFY_OK;
+static int virtnet_cpu_down_prep(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+
+	virtnet_clean_affinity(vi, cpu);
+	return 0;
+}
+
+static enum cpuhp_state virtionet_online;
+
+static int virtnet_cpu_notif_add(struct virtnet_info *vi)
+{
+	int ret;
+
+	ret = cpuhp_state_add_instance_nocalls(virtionet_online, &vi->node);
+	if (ret)
+		return ret;
+	ret = cpuhp_state_add_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					       &vi->node_dead);
+	if (!ret)
+		return ret;
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	return ret;
+}
+
+static void virtnet_cpu_notif_remove(struct virtnet_info *vi)
+{
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	cpuhp_state_remove_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					    &vi->node_dead);
 }
 
 static void virtnet_get_ringparam(struct net_device *dev,
@@ -1879,8 +1908,7 @@ static int virtnet_probe(struct virtio_device *vdev)
 
 	virtio_device_ready(vdev);
 
-	vi->nb.notifier_call = &virtnet_cpu_callback;
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err) {
 		pr_debug("virtio_net: registering cpu notifier failed\n");
 		goto free_unregister_netdev;
@@ -1934,7 +1962,7 @@ static void virtnet_remove(struct virtio_device *vdev)
 {
 	struct virtnet_info *vi = vdev->priv;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device. */
 	flush_work(&vi->config_work);
@@ -1953,7 +1981,7 @@ static int virtnet_freeze(struct virtio_device *vdev)
 	struct virtnet_info *vi = vdev->priv;
 	int i;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device */
 	flush_work(&vi->config_work);
@@ -1997,7 +2025,7 @@ static int virtnet_restore(struct virtio_device *vdev)
 	virtnet_set_queues(vi, vi->curr_queue_pairs);
 	rtnl_unlock();
 
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err)
 		return err;
 
@@ -2039,7 +2067,41 @@ static struct virtio_driver virtio_net_driver = {
 #endif
 };
 
-module_virtio_driver(virtio_net_driver);
+static __init int virtio_net_driver_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "AP_VIRT_NET_ONLINE",
+				      virtnet_cpu_online,
+				      virtnet_cpu_down_prep);
+	if (ret < 0)
+		goto out;
+	virtionet_online = ret;
+	ret = cpuhp_setup_state_multi(CPUHP_VIRT_NET_DEAD, "VIRT_NET_DEAD",
+				      NULL, virtnet_cpu_dead);
+	if (ret)
+		goto err_dead;
+
+        ret = register_virtio_driver(&virtio_net_driver);
+	if (ret)
+		goto err_virtio;
+	return 0;
+err_virtio:
+	cpuhp_remove_state_nocalls(CPUHP_VIRT_NET_DEAD);
+err_dead:
+	cpuhp_remove_state_nocalls(virtionet_online);
+out:
+	return ret;
+}
+module_init(virtio_net_driver_init);
+
+static __exit void virtio_net_driver_exit(void)
+{
+	cpuhp_remove_state_nocalls(CPUHP_VIRT_NET_DEAD);
+	cpuhp_remove_state_nocalls(virtionet_online);
+	unregister_virtio_driver(&virtio_net_driver);
+}
+module_exit(virtio_net_driver_exit);
 
 MODULE_DEVICE_TABLE(virtio, id_table);
 MODULE_DESCRIPTION("Virtio network driver");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 81cc20789a3e..5c657c836de6 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,6 +20,7 @@ enum cpuhp_state {
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
+	CPUHP_VIRT_NET_DEAD,
 	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_TIMERS_DEAD,
-- 
2.8.1

[tip:smp/hotplug] net/virtio-net: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  6dc465457fac9a49462bebbb870beea5d536b7c2
Gitweb:     http://git.kernel.org/tip/6dc465457fac9a49462bebbb870beea5d536b7c2
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:43 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Thu, 1 Sep 2016 18:38:29 +0200

net/virtio-net: Convert to hotplug state machine

Install the callbacks via the state machine.  The driver supports multiple
instances and therefore the new cpuhp_state_add_instance_nocalls()
infrastrucure is used. The driver currently uses get_online_cpus() to avoid
missing a CPU hotplug event while invoking virtnet_set_affinity(). This could
be avoided by using cpuhp_state_add_instance() variant which holds the hotplug
lock and invokes callback during registration. This is more or less a 1:1
conversion of the current code.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: netdev@vger.kernel.org
Cc: Will Deacon <will.deacon@arm.com>
Cc: virtualization@lists.linux-foundation.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/1471024183-12666-7-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/net/virtio_net.c   | 110 +++++++++++++++++++++++++++++++++++----------
 include/linux/cpuhotplug.h |   1 +
 2 files changed, 87 insertions(+), 24 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 1b5f531..e9be88c 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -138,8 +138,9 @@ struct virtnet_info {
 	/* Does the affinity hint is set for virtqueues? */
 	bool affinity_hint_set;
 
-	/* CPU hot plug notifier */
-	struct notifier_block nb;
+	/* CPU hotplug instances for online & dead */
+	struct hlist_node node;
+	struct hlist_node node_dead;
 
 	/* Control VQ buffers: protected by the rtnl lock */
 	struct virtio_net_ctrl_hdr ctrl_hdr;
@@ -1237,25 +1238,53 @@ static void virtnet_set_affinity(struct virtnet_info *vi)
 	vi->affinity_hint_set = true;
 }
 
-static int virtnet_cpu_callback(struct notifier_block *nfb,
-			        unsigned long action, void *hcpu)
+static int virtnet_cpu_online(unsigned int cpu, struct hlist_node *node)
 {
-	struct virtnet_info *vi = container_of(nfb, struct virtnet_info, nb);
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	switch(action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-	case CPU_DEAD:
-		virtnet_set_affinity(vi);
-		break;
-	case CPU_DOWN_PREPARE:
-		virtnet_clean_affinity(vi, (long)hcpu);
-		break;
-	default:
-		break;
-	}
+static int virtnet_cpu_dead(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node_dead);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	return NOTIFY_OK;
+static int virtnet_cpu_down_prep(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+
+	virtnet_clean_affinity(vi, cpu);
+	return 0;
+}
+
+static enum cpuhp_state virtionet_online;
+
+static int virtnet_cpu_notif_add(struct virtnet_info *vi)
+{
+	int ret;
+
+	ret = cpuhp_state_add_instance_nocalls(virtionet_online, &vi->node);
+	if (ret)
+		return ret;
+	ret = cpuhp_state_add_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					       &vi->node_dead);
+	if (!ret)
+		return ret;
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	return ret;
+}
+
+static void virtnet_cpu_notif_remove(struct virtnet_info *vi)
+{
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	cpuhp_state_remove_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					    &vi->node_dead);
 }
 
 static void virtnet_get_ringparam(struct net_device *dev,
@@ -1879,8 +1908,7 @@ static int virtnet_probe(struct virtio_device *vdev)
 
 	virtio_device_ready(vdev);
 
-	vi->nb.notifier_call = &virtnet_cpu_callback;
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err) {
 		pr_debug("virtio_net: registering cpu notifier failed\n");
 		goto free_unregister_netdev;
@@ -1934,7 +1962,7 @@ static void virtnet_remove(struct virtio_device *vdev)
 {
 	struct virtnet_info *vi = vdev->priv;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device. */
 	flush_work(&vi->config_work);
@@ -1953,7 +1981,7 @@ static int virtnet_freeze(struct virtio_device *vdev)
 	struct virtnet_info *vi = vdev->priv;
 	int i;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device */
 	flush_work(&vi->config_work);
@@ -1997,7 +2025,7 @@ static int virtnet_restore(struct virtio_device *vdev)
 	virtnet_set_queues(vi, vi->curr_queue_pairs);
 	rtnl_unlock();
 
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err)
 		return err;
 
@@ -2039,7 +2067,41 @@ static struct virtio_driver virtio_net_driver = {
 #endif
 };
 
-module_virtio_driver(virtio_net_driver);
+static __init int virtio_net_driver_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "AP_VIRT_NET_ONLINE",
+				      virtnet_cpu_online,
+				      virtnet_cpu_down_prep);
+	if (ret < 0)
+		goto out;
+	virtionet_online = ret;
+	ret = cpuhp_setup_state_multi(CPUHP_VIRT_NET_DEAD, "VIRT_NET_DEAD",
+				      NULL, virtnet_cpu_dead);
+	if (ret)
+		goto err_dead;
+
+        ret = register_virtio_driver(&virtio_net_driver);
+	if (ret)
+		goto err_virtio;
+	return 0;
+err_virtio:
+	cpuhp_remove_state_nocalls(CPUHP_VIRT_NET_DEAD);
+err_dead:
+	cpuhp_remove_state_nocalls(virtionet_online);
+out:
+	return ret;
+}
+module_init(virtio_net_driver_init);
+
+static __exit void virtio_net_driver_exit(void)
+{
+	cpuhp_remove_state_nocalls(CPUHP_VIRT_NET_DEAD);
+	cpuhp_remove_state_nocalls(virtionet_online);
+	unregister_virtio_driver(&virtio_net_driver);
+}
+module_exit(virtio_net_driver_exit);
 
 MODULE_DEVICE_TABLE(virtio, id_table);
 MODULE_DESCRIPTION("Virtio network driver");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 81cc207..5c657c8 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -20,6 +20,7 @@ enum cpuhp_state {
 	CPUHP_PROFILE_PREPARE,
 	CPUHP_X2APIC_PREPARE,
 	CPUHP_SMPCFD_PREPARE,
+	CPUHP_VIRT_NET_DEAD,
 	CPUHP_RCUTREE_PREP,
 	CPUHP_NOTIFY_PREPARE,
 	CPUHP_TIMERS_DEAD,

[tip:smp/hotplug] net/virtio-net: Convert to hotplug state machine

[tip-bot for Sebastian Andrzej Siewior]

Commit-ID:  8017c279196ab29174bafc104ac4ebbd42c7ca7f
Gitweb:     http://git.kernel.org/tip/8017c279196ab29174bafc104ac4ebbd42c7ca7f
Author:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
AuthorDate: Fri, 12 Aug 2016 19:49:43 +0200
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Fri, 2 Sep 2016 20:05:06 +0200

net/virtio-net: Convert to hotplug state machine

Install the callbacks via the state machine.

The driver supports multiple instances and therefore the new
cpuhp_state_add_instance_nocalls() infrastrucure is used. The driver
currently uses get_online_cpus() to avoid missing a CPU hotplug event while
invoking virtnet_set_affinity(). This could be avoided by using
cpuhp_state_add_instance() variant which holds the hotplug lock and invokes
callback during registration. This is more or less a 1:1 conversion of the
current code.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: netdev@vger.kernel.org
Cc: Will Deacon <will.deacon@arm.com>
Cc: virtualization@lists.linux-foundation.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/1471024183-12666-7-git-send-email-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 drivers/net/virtio_net.c   | 110 +++++++++++++++++++++++++++++++++++----------
 include/linux/cpuhotplug.h |   1 +
 2 files changed, 87 insertions(+), 24 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 1b5f531..fad84f3 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -138,8 +138,9 @@ struct virtnet_info {
 	/* Does the affinity hint is set for virtqueues? */
 	bool affinity_hint_set;
 
-	/* CPU hot plug notifier */
-	struct notifier_block nb;
+	/* CPU hotplug instances for online & dead */
+	struct hlist_node node;
+	struct hlist_node node_dead;
 
 	/* Control VQ buffers: protected by the rtnl lock */
 	struct virtio_net_ctrl_hdr ctrl_hdr;
@@ -1237,25 +1238,53 @@ static void virtnet_set_affinity(struct virtnet_info *vi)
 	vi->affinity_hint_set = true;
 }
 
-static int virtnet_cpu_callback(struct notifier_block *nfb,
-			        unsigned long action, void *hcpu)
+static int virtnet_cpu_online(unsigned int cpu, struct hlist_node *node)
 {
-	struct virtnet_info *vi = container_of(nfb, struct virtnet_info, nb);
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	switch(action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-	case CPU_DOWN_FAILED:
-	case CPU_DEAD:
-		virtnet_set_affinity(vi);
-		break;
-	case CPU_DOWN_PREPARE:
-		virtnet_clean_affinity(vi, (long)hcpu);
-		break;
-	default:
-		break;
-	}
+static int virtnet_cpu_dead(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node_dead);
+	virtnet_set_affinity(vi);
+	return 0;
+}
 
-	return NOTIFY_OK;
+static int virtnet_cpu_down_prep(unsigned int cpu, struct hlist_node *node)
+{
+	struct virtnet_info *vi = hlist_entry_safe(node, struct virtnet_info,
+						   node);
+
+	virtnet_clean_affinity(vi, cpu);
+	return 0;
+}
+
+static enum cpuhp_state virtionet_online;
+
+static int virtnet_cpu_notif_add(struct virtnet_info *vi)
+{
+	int ret;
+
+	ret = cpuhp_state_add_instance_nocalls(virtionet_online, &vi->node);
+	if (ret)
+		return ret;
+	ret = cpuhp_state_add_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					       &vi->node_dead);
+	if (!ret)
+		return ret;
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	return ret;
+}
+
+static void virtnet_cpu_notif_remove(struct virtnet_info *vi)
+{
+	cpuhp_state_remove_instance_nocalls(virtionet_online, &vi->node);
+	cpuhp_state_remove_instance_nocalls(CPUHP_VIRT_NET_DEAD,
+					    &vi->node_dead);
 }
 
 static void virtnet_get_ringparam(struct net_device *dev,
@@ -1879,8 +1908,7 @@ static int virtnet_probe(struct virtio_device *vdev)
 
 	virtio_device_ready(vdev);
 
-	vi->nb.notifier_call = &virtnet_cpu_callback;
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err) {
 		pr_debug("virtio_net: registering cpu notifier failed\n");
 		goto free_unregister_netdev;
@@ -1934,7 +1962,7 @@ static void virtnet_remove(struct virtio_device *vdev)
 {
 	struct virtnet_info *vi = vdev->priv;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device. */
 	flush_work(&vi->config_work);
@@ -1953,7 +1981,7 @@ static int virtnet_freeze(struct virtio_device *vdev)
 	struct virtnet_info *vi = vdev->priv;
 	int i;
 
-	unregister_hotcpu_notifier(&vi->nb);
+	virtnet_cpu_notif_remove(vi);
 
 	/* Make sure no work handler is accessing the device */
 	flush_work(&vi->config_work);
@@ -1997,7 +2025,7 @@ static int virtnet_restore(struct virtio_device *vdev)
 	virtnet_set_queues(vi, vi->curr_queue_pairs);
 	rtnl_unlock();
 
-	err = register_hotcpu_notifier(&vi->nb);
+	err = virtnet_cpu_notif_add(vi);
 	if (err)
 		return err;
 
@@ -2039,7 +2067,41 @@ static struct virtio_driver virtio_net_driver = {
 #endif
 };
 
-module_virtio_driver(virtio_net_driver);
+static __init int virtio_net_driver_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "AP_VIRT_NET_ONLINE",
+				      virtnet_cpu_online,
+				      virtnet_cpu_down_prep);
+	if (ret < 0)
+		goto out;
+	virtionet_online = ret;
+	ret = cpuhp_setup_state_multi(CPUHP_VIRT_NET_DEAD, "VIRT_NET_DEAD",
+				      NULL, virtnet_cpu_dead);
+	if (ret)
+		goto err_dead;
+
+        ret = register_virtio_driver(&virtio_net_driver);
+	if (ret)
+		goto err_virtio;
+	return 0;
+err_virtio:
+	cpuhp_remove_multi_state(CPUHP_VIRT_NET_DEAD);
+err_dead:
+	cpuhp_remove_multi_state(virtionet_online);
+out:
+	return ret;
+}
+module_init(virtio_net_driver_init);
+
+static __exit void virtio_net_driver_exit(void)
+{
+	cpuhp_remove_multi_state(CPUHP_VIRT_NET_DEAD);
+	cpuhp_remove_multi_state(virtionet_online);
+	unregister_virtio_driver(&virtio_net_driver);
+}
+module_exit(virtio_net_driver_exit);
 
 MODULE_DEVICE_TABLE(virtio, id_table);
 MODULE_DESCRIPTION("Virtio network driver");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index dcfe6191..b95f7ad 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_X86_APB_DEAD,
+	CPUHP_VIRT_NET_DEAD,
 	CPUHP_WORKQUEUE_PREP,
 	CPUHP_POWER_NUMA_PREPARE,
 	CPUHP_HRTIMERS_PREPARE,