Merge branch 'core' of git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile into oprofile

3 files changed, 130 insertions, 113 deletions

diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index 9f001d9..b28d2f1 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -31,8 +31,9 @@ static struct op_x86_model_spec *model;
 static DEFINE_PER_CPU(struct op_msrs, cpu_msrs);
 static DEFINE_PER_CPU(unsigned long, saved_lvtpc);
 
-/* 0 == registered but off, 1 == registered and on */
-static int nmi_enabled = 0;
+/* must be protected with get_online_cpus()/put_online_cpus(): */
+static int nmi_enabled;
+static int ctr_running;
 
 struct op_counter_config counter_config[OP_MAX_COUNTER];
 
@@ -61,12 +62,16 @@ static int profile_exceptions_notify(struct notifier_block *self,
 {
 	struct die_args *args = (struct die_args *)data;
 	int ret = NOTIFY_DONE;
-	int cpu = smp_processor_id();
 
 	switch (val) {
 	case DIE_NMI:
 	case DIE_NMI_IPI:
-		model->check_ctrs(args->regs, &per_cpu(cpu_msrs, cpu));
+		if (ctr_running)
+			model->check_ctrs(args->regs, &__get_cpu_var(cpu_msrs));
+		else if (!nmi_enabled)
+			break;
+		else
+			model->stop(&__get_cpu_var(cpu_msrs));
 		ret = NOTIFY_STOP;
 		break;
 	default:
@@ -95,24 +100,36 @@ static void nmi_cpu_save_registers(struct op_msrs *msrs)
 static void nmi_cpu_start(void *dummy)
 {
 	struct op_msrs const *msrs = &__get_cpu_var(cpu_msrs);
-	model->start(msrs);
+	if (!msrs->controls)
+		WARN_ON_ONCE(1);
+	else
+		model->start(msrs);
 }
 
 static int nmi_start(void)
 {
+	get_online_cpus();
 	on_each_cpu(nmi_cpu_start, NULL, 1);
+	ctr_running = 1;
+	put_online_cpus();
 	return 0;
 }
 
 static void nmi_cpu_stop(void *dummy)
 {
 	struct op_msrs const *msrs = &__get_cpu_var(cpu_msrs);
-	model->stop(msrs);
+	if (!msrs->controls)
+		WARN_ON_ONCE(1);
+	else
+		model->stop(msrs);
 }
 
 static void nmi_stop(void)
 {
+	get_online_cpus();
 	on_each_cpu(nmi_cpu_stop, NULL, 1);
+	ctr_running = 0;
+	put_online_cpus();
 }
 
 #ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
@@ -252,7 +269,10 @@ static int nmi_switch_event(void)
 	if (nmi_multiplex_on() < 0)
 		return -EINVAL;		/* not necessary */
 
-	on_each_cpu(nmi_cpu_switch, NULL, 1);
+	get_online_cpus();
+	if (ctr_running)
+		on_each_cpu(nmi_cpu_switch, NULL, 1);
+	put_online_cpus();
 
 	return 0;
 }
@@ -344,50 +364,6 @@ static struct notifier_block profile_exceptions_nb = {
 	.priority = 2
 };
 
-static int nmi_setup(void)
-{
-	int err = 0;
-	int cpu;
-
-	if (!allocate_msrs())
-		return -ENOMEM;
-
-	/* We need to serialize save and setup for HT because the subset
-	 * of msrs are distinct for save and setup operations
-	 */
-
-	/* Assume saved/restored counters are the same on all CPUs */
-	err = model->fill_in_addresses(&per_cpu(cpu_msrs, 0));
-	if (err)
-		goto fail;
-
-	for_each_possible_cpu(cpu) {
-		if (!cpu)
-			continue;
-
-		memcpy(per_cpu(cpu_msrs, cpu).counters,
-		       per_cpu(cpu_msrs, 0).counters,
-		       sizeof(struct op_msr) * model->num_counters);
-
-		memcpy(per_cpu(cpu_msrs, cpu).controls,
-		       per_cpu(cpu_msrs, 0).controls,
-		       sizeof(struct op_msr) * model->num_controls);
-
-		mux_clone(cpu);
-	}
-
-	err = register_die_notifier(&profile_exceptions_nb);
-	if (err)
-		goto fail;
-
-	on_each_cpu(nmi_cpu_setup, NULL, 1);
-	nmi_enabled = 1;
-	return 0;
-fail:
-	free_msrs();
-	return err;
-}
-
 static void nmi_cpu_restore_registers(struct op_msrs *msrs)
 {
 	struct op_msr *counters = msrs->counters;
@@ -421,19 +397,24 @@ static void nmi_cpu_shutdown(void *dummy)
 	apic_write(APIC_LVTPC, per_cpu(saved_lvtpc, cpu));
 	apic_write(APIC_LVTERR, v);
 	nmi_cpu_restore_registers(msrs);
+	if (model->cpu_down)
+		model->cpu_down();
 }
 
-static void nmi_shutdown(void)
+static void nmi_cpu_up(void *dummy)
 {
-	struct op_msrs *msrs;
+	if (nmi_enabled)
+		nmi_cpu_setup(dummy);
+	if (ctr_running)
+		nmi_cpu_start(dummy);
+}
 
-	nmi_enabled = 0;
-	on_each_cpu(nmi_cpu_shutdown, NULL, 1);
-	unregister_die_notifier(&profile_exceptions_nb);
-	msrs = &get_cpu_var(cpu_msrs);
-	model->shutdown(msrs);
-	free_msrs();
-	put_cpu_var(cpu_msrs);
+static void nmi_cpu_down(void *dummy)
+{
+	if (ctr_running)
+		nmi_cpu_stop(dummy);
+	if (nmi_enabled)
+		nmi_cpu_shutdown(dummy);
 }
 
 static int nmi_create_files(struct super_block *sb, struct dentry *root)
@@ -465,7 +446,6 @@ static int nmi_create_files(struct super_block *sb, struct dentry *root)
 	return 0;
 }
 
-#ifdef CONFIG_SMP
 static int oprofile_cpu_notifier(struct notifier_block *b, unsigned long action,
 				 void *data)
 {
@@ -473,10 +453,10 @@ static int oprofile_cpu_notifier(struct notifier_block *b, unsigned long action,
 	switch (action) {
 	case CPU_DOWN_FAILED:
 	case CPU_ONLINE:
-		smp_call_function_single(cpu, nmi_cpu_start, NULL, 0);
+		smp_call_function_single(cpu, nmi_cpu_up, NULL, 0);
 		break;
 	case CPU_DOWN_PREPARE:
-		smp_call_function_single(cpu, nmi_cpu_stop, NULL, 1);
+		smp_call_function_single(cpu, nmi_cpu_down, NULL, 1);
 		break;
 	}
 	return NOTIFY_DONE;
@@ -485,7 +465,75 @@ static int oprofile_cpu_notifier(struct notifier_block *b, unsigned long action,
 static struct notifier_block oprofile_cpu_nb = {
 	.notifier_call = oprofile_cpu_notifier
 };
-#endif
+
+static int nmi_setup(void)
+{
+	int err = 0;
+	int cpu;
+
+	if (!allocate_msrs())
+		return -ENOMEM;
+
+	/* We need to serialize save and setup for HT because the subset
+	 * of msrs are distinct for save and setup operations
+	 */
+
+	/* Assume saved/restored counters are the same on all CPUs */
+	err = model->fill_in_addresses(&per_cpu(cpu_msrs, 0));
+	if (err)
+		goto fail;
+
+	for_each_possible_cpu(cpu) {
+		if (!cpu)
+			continue;
+
+		memcpy(per_cpu(cpu_msrs, cpu).counters,
+		       per_cpu(cpu_msrs, 0).counters,
+		       sizeof(struct op_msr) * model->num_counters);
+
+		memcpy(per_cpu(cpu_msrs, cpu).controls,
+		       per_cpu(cpu_msrs, 0).controls,
+		       sizeof(struct op_msr) * model->num_controls);
+
+		mux_clone(cpu);
+	}
+
+	nmi_enabled = 0;
+	ctr_running = 0;
+	barrier();
+	err = register_die_notifier(&profile_exceptions_nb);
+	if (err)
+		goto fail;
+
+	get_online_cpus();
+	register_cpu_notifier(&oprofile_cpu_nb);
+	on_each_cpu(nmi_cpu_setup, NULL, 1);
+	nmi_enabled = 1;
+	put_online_cpus();
+
+	return 0;
+fail:
+	free_msrs();
+	return err;
+}
+
+static void nmi_shutdown(void)
+{
+	struct op_msrs *msrs;
+
+	get_online_cpus();
+	unregister_cpu_notifier(&oprofile_cpu_nb);
+	on_each_cpu(nmi_cpu_shutdown, NULL, 1);
+	nmi_enabled = 0;
+	ctr_running = 0;
+	put_online_cpus();
+	barrier();
+	unregister_die_notifier(&profile_exceptions_nb);
+	msrs = &get_cpu_var(cpu_msrs);
+	model->shutdown(msrs);
+	free_msrs();
+	put_cpu_var(cpu_msrs);
+}
 
 #ifdef CONFIG_PM
 
@@ -695,9 +743,6 @@ int __init op_nmi_init(struct oprofile_operations *ops)
 		return -ENODEV;
 	}
 
-#ifdef CONFIG_SMP
-	register_cpu_notifier(&oprofile_cpu_nb);
-#endif
 	/* default values, can be overwritten by model */
 	ops->create_files	= nmi_create_files;
 	ops->setup		= nmi_setup;
@@ -724,12 +769,6 @@ int __init op_nmi_init(struct oprofile_operations *ops)
 
 void op_nmi_exit(void)
 {
-	if (using_nmi) {
+	if (using_nmi)
 		exit_sysfs();
-#ifdef CONFIG_SMP
-		unregister_cpu_notifier(&oprofile_cpu_nb);
-#endif
-	}
-	if (model->exit)
-		model->exit();
 }
diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 384c524..b67a6b5 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -374,6 +374,15 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 		val |= op_x86_get_ctrl(model, &counter_config[virt]);
 		wrmsrl(msrs->controls[i].addr, val);
 	}
+
+	if (ibs_caps)
+		setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_NMI, 0);
+}
+
+static void op_amd_cpu_shutdown(void)
+{
+	if (ibs_caps)
+		setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_FIX, 1);
 }
 
 static int op_amd_check_ctrs(struct pt_regs * const regs,
@@ -436,28 +445,16 @@ static void op_amd_stop(struct op_msrs const * const msrs)
 	op_amd_stop_ibs();
 }
 
-static u8 ibs_eilvt_off;
-
-static inline void apic_init_ibs_nmi_per_cpu(void *arg)
-{
-	ibs_eilvt_off = setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_NMI, 0);
-}
-
-static inline void apic_clear_ibs_nmi_per_cpu(void *arg)
-{
-	setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_FIX, 1);
-}
-
-static int init_ibs_nmi(void)
+static int __init_ibs_nmi(void)
 {
 #define IBSCTL_LVTOFFSETVAL		(1 << 8)
 #define IBSCTL				0x1cc
 	struct pci_dev *cpu_cfg;
 	int nodes;
 	u32 value = 0;
+	u8 ibs_eilvt_off;
 
-	/* per CPU setup */
-	on_each_cpu(apic_init_ibs_nmi_per_cpu, NULL, 1);
+	ibs_eilvt_off = setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_FIX, 1);
 
 	nodes = 0;
 	cpu_cfg = NULL;
@@ -487,21 +484,15 @@ static int init_ibs_nmi(void)
 	return 0;
 }
 
-/* uninitialize the APIC for the IBS interrupts if needed */
-static void clear_ibs_nmi(void)
-{
-	on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1);
-}
-
 /* initialize the APIC for the IBS interrupts if available */
-static void ibs_init(void)
+static void init_ibs(void)
 {
 	ibs_caps = get_ibs_caps();
 
 	if (!ibs_caps)
 		return;
 
-	if (init_ibs_nmi()) {
+	if (__init_ibs_nmi()) {
 		ibs_caps = 0;
 		return;
 	}
@@ -510,14 +501,6 @@ static void ibs_init(void)
 	       (unsigned)ibs_caps);
 }
 
-static void ibs_exit(void)
-{
-	if (!ibs_caps)
-		return;
-
-	clear_ibs_nmi();
-}
-
 static int (*create_arch_files)(struct super_block *sb, struct dentry *root);
 
 static int setup_ibs_files(struct super_block *sb, struct dentry *root)
@@ -566,17 +549,12 @@ static int setup_ibs_files(struct super_block *sb, struct dentry *root)
 
 static int op_amd_init(struct oprofile_operations *ops)
 {
-	ibs_init();
+	init_ibs();
 	create_arch_files = ops->create_files;
 	ops->create_files = setup_ibs_files;
 	return 0;
 }
 
-static void op_amd_exit(void)
-{
-	ibs_exit();
-}
-
 struct op_x86_model_spec op_amd_spec = {
 	.num_counters		= NUM_COUNTERS,
 	.num_controls		= NUM_COUNTERS,
@@ -584,9 +562,9 @@ struct op_x86_model_spec op_amd_spec = {
 	.reserved		= MSR_AMD_EVENTSEL_RESERVED,
 	.event_mask		= OP_EVENT_MASK,
 	.init			= op_amd_init,
-	.exit			= op_amd_exit,
 	.fill_in_addresses	= &op_amd_fill_in_addresses,
 	.setup_ctrs		= &op_amd_setup_ctrs,
+	.cpu_down		= &op_amd_cpu_shutdown,
 	.check_ctrs		= &op_amd_check_ctrs,
 	.start			= &op_amd_start,
 	.stop			= &op_amd_stop,
diff --git a/arch/x86/oprofile/op_x86_model.h b/arch/x86/oprofile/op_x86_model.h
index 5514013..89017fa 100644
--- a/arch/x86/oprofile/op_x86_model.h
+++ b/arch/x86/oprofile/op_x86_model.h
@@ -40,10 +40,10 @@ struct op_x86_model_spec {
 	u64		reserved;
 	u16		event_mask;
 	int		(*init)(struct oprofile_operations *ops);
-	void		(*exit)(void);
 	int		(*fill_in_addresses)(struct op_msrs * const msrs);
 	void		(*setup_ctrs)(struct op_x86_model_spec const *model,
 				      struct op_msrs const * const msrs);
+	void		(*cpu_down)(void);
 	int		(*check_ctrs)(struct pt_regs * const regs,
 				      struct op_msrs const * const msrs);
 	void		(*start)(struct op_msrs const * const msrs);