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 <[e-mail]>
Signed-off-by: Sebastian Andrzej Siewior <[e-mail]>
Cc: Mark Rutland <[e-mail]>
Cc: Peter Zijlstra <[e-mail]>
Cc: Will Deacon <[e-mail]>
Cc: [e-mail]
Link: http://lkml.kernel.org/r/[e-mail]
Signed-off-by: Thomas Gleixner <[e-mail]>

include/linux/cpuhotplug.h

···
 
 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
···
 				    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);
 }
 
 /**
···
 					    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);
···
 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

include/trace/events/cpuhp.h

···
 		  __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,

kernel/cpu.c

···
 	bool			rollback;
 	bool			single;
 	bool			bringup;
+	struct hlist_node	*node;
 	enum cpuhp_state	cb_state;
 	int			result;
 	struct completion	done;
···
  * @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);
···
  * @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;
 }
···
 		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);
 	}
 }
 
···
 	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);
···
 		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);
 	}
 }
 
···
 
 	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);
···
 		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);
···
 
 /* 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);
 
···
 	 * 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
···
 	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();
···
 
 	while (st->state < target) {
 		st->state++;
-		cpuhp_invoke_callback(cpu, st->state, true);
+		cpuhp_invoke_callback(cpu, st->state, true, NULL);
 	}
 }
 
···
 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;
···
 	sp->startup = startup;
 	sp->teardown = teardown;
 	sp->name = name;
+	sp->multi_instance = multi_instance;
+	INIT_HLIST_HEAD(&sp->list);
 	mutex_unlock(&cpuhp_state_mutex);
 }
 
···
  * 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;
···
 	 */
 #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;
···
  *
  * 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;
 
···
 
 		/* 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);
 	}
 }
 
···
 	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
···
 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;
···
 		state = ret;
 	}
 
-	cpuhp_store_callbacks(state, name, startup, teardown);
+	cpuhp_store_callbacks(state, name, startup, teardown, multi_instance);
 
 	if (!invoke || !startup)
 		goto out;
···
 		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;
 		}
 	}
···
 }
 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
···
  */
 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;
···
 		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);