oprofile: Implement performance counter multiplexing

The number of hardware counters is limited. The multiplexing feature
enables OProfile to gather more events than counters are provided by
the hardware. This is realized by switching between events at an user
specified time interval.

A new file (/dev/oprofile/time_slice) is added for the user to specify
the timer interval in ms. If the number of events to profile is higher
than the number of hardware counters available, the patch will
schedule a work queue that switches the event counter and re-writes
the different sets of values into it. The switching mechanism needs to
be implemented for each architecture to support multiplexing. This
patch only implements AMD CPU support, but multiplexing can be easily
extended for other models and architectures.

There are follow-on patches that rework parts of this patch.

Signed-off-by: Jason Yeh <jason.yeh@amd.com>
Signed-off-by: Robert Richter <robert.richter@amd.com>

6 files changed, 267 insertions, 20 deletions

diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index fca8dc94531e..e54f6a0b35ac 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -1,11 +1,14 @@
 /**
  * @file nmi_int.c
  *
- * @remark Copyright 2002-2008 OProfile authors
+ * @remark Copyright 2002-2009 OProfile authors
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
  * @author Robert Richter <robert.richter@amd.com>
+ * @author Barry Kasindorf <barry.kasindorf@amd.com>
+ * @author Jason Yeh <jason.yeh@amd.com>
+ * @author Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
  */
 
 #include <linux/init.h>
@@ -24,6 +27,12 @@
 #include "op_counter.h"
 #include "op_x86_model.h"
 
+
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+DEFINE_PER_CPU(int, switch_index);
+#endif
+
+
 static struct op_x86_model_spec const *model;
 static DEFINE_PER_CPU(struct op_msrs, cpu_msrs);
 static DEFINE_PER_CPU(unsigned long, saved_lvtpc);
@@ -31,6 +40,13 @@ static DEFINE_PER_CPU(unsigned long, saved_lvtpc);
 /* 0 == registered but off, 1 == registered and on */
 static int nmi_enabled = 0;
 
+
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+extern atomic_t multiplex_counter;
+#endif
+
+struct op_counter_config counter_config[OP_MAX_COUNTER];
+
 /* common functions */
 
 u64 op_x86_get_ctrl(struct op_x86_model_spec const *model,
@@ -95,6 +111,11 @@ static void free_msrs(void)
 		per_cpu(cpu_msrs, i).counters = NULL;
 		kfree(per_cpu(cpu_msrs, i).controls);
 		per_cpu(cpu_msrs, i).controls = NULL;
+
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		kfree(per_cpu(cpu_msrs, i).multiplex);
+		per_cpu(cpu_msrs, i).multiplex = NULL;
+#endif
 	}
 }
 
@@ -103,6 +124,9 @@ static int allocate_msrs(void)
 	int success = 1;
 	size_t controls_size = sizeof(struct op_msr) * model->num_controls;
 	size_t counters_size = sizeof(struct op_msr) * model->num_counters;
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	size_t multiplex_size = sizeof(struct op_msr) * model->num_virt_counters;
+#endif
 
 	int i;
 	for_each_possible_cpu(i) {
@@ -118,6 +142,14 @@ static int allocate_msrs(void)
 			success = 0;
 			break;
 		}
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		per_cpu(cpu_msrs, i).multiplex =
+				kmalloc(multiplex_size, GFP_KERNEL);
+		if (!per_cpu(cpu_msrs, i).multiplex) {
+			success = 0;
+			break;
+		}
+#endif
 	}
 
 	if (!success)
@@ -126,6 +158,25 @@ static int allocate_msrs(void)
 	return success;
 }
 
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void nmi_setup_cpu_mux(struct op_msrs const * const msrs)
+{
+	int i;
+	struct op_msr *multiplex = msrs->multiplex;
+
+	for (i = 0; i < model->num_virt_counters; ++i) {
+		if (counter_config[i].enabled) {
+			multiplex[i].saved = -(u64)counter_config[i].count;
+		} else {
+			multiplex[i].addr  = 0;
+			multiplex[i].saved = 0;
+		}
+	}
+}
+
+#endif
+
 static void nmi_cpu_setup(void *dummy)
 {
 	int cpu = smp_processor_id();
@@ -133,6 +184,9 @@ static void nmi_cpu_setup(void *dummy)
 	nmi_cpu_save_registers(msrs);
 	spin_lock(&oprofilefs_lock);
 	model->setup_ctrs(model, msrs);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	nmi_setup_cpu_mux(msrs);
+#endif
 	spin_unlock(&oprofilefs_lock);
 	per_cpu(saved_lvtpc, cpu) = apic_read(APIC_LVTPC);
 	apic_write(APIC_LVTPC, APIC_DM_NMI);
@@ -173,14 +227,52 @@ static int nmi_setup(void)
 			memcpy(per_cpu(cpu_msrs, cpu).controls,
 				per_cpu(cpu_msrs, 0).controls,
 				sizeof(struct op_msr) * model->num_controls);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+			memcpy(per_cpu(cpu_msrs, cpu).multiplex,
+				per_cpu(cpu_msrs, 0).multiplex,
+				sizeof(struct op_msr) * model->num_virt_counters);
+#endif
 		}
-
 	}
 	on_each_cpu(nmi_cpu_setup, NULL, 1);
 	nmi_enabled = 1;
 	return 0;
 }
 
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void nmi_cpu_save_mpx_registers(struct op_msrs *msrs)
+{
+	unsigned int si = __get_cpu_var(switch_index);
+	struct op_msr *multiplex = msrs->multiplex;
+	unsigned int i;
+
+	for (i = 0; i < model->num_counters; ++i) {
+		int offset = i + si;
+		if (multiplex[offset].addr) {
+			rdmsrl(multiplex[offset].addr,
+			       multiplex[offset].saved);
+		}
+	}
+}
+
+static void nmi_cpu_restore_mpx_registers(struct op_msrs *msrs)
+{
+	unsigned int si = __get_cpu_var(switch_index);
+	struct op_msr *multiplex = msrs->multiplex;
+	unsigned int i;
+
+	for (i = 0; i < model->num_counters; ++i) {
+		int offset = i + si;
+		if (multiplex[offset].addr) {
+			wrmsrl(multiplex[offset].addr,
+			       multiplex[offset].saved);
+		}
+	}
+}
+
+#endif
+
 static void nmi_cpu_restore_registers(struct op_msrs *msrs)
 {
 	struct op_msr *counters = msrs->counters;
@@ -214,6 +306,9 @@ 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);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	__get_cpu_var(switch_index) = 0;
+#endif
 }
 
 static void nmi_shutdown(void)
@@ -252,16 +347,15 @@ static void nmi_stop(void)
 	on_each_cpu(nmi_cpu_stop, NULL, 1);
 }
 
-struct op_counter_config counter_config[OP_MAX_COUNTER];
-
 static int nmi_create_files(struct super_block *sb, struct dentry *root)
 {
 	unsigned int i;
 
-	for (i = 0; i < model->num_counters; ++i) {
+	for (i = 0; i < model->num_virt_counters; ++i) {
 		struct dentry *dir;
 		char buf[4];
 
+#ifndef CONFIG_OPROFILE_EVENT_MULTIPLEX
 		/* quick little hack to _not_ expose a counter if it is not
 		 * available for use.  This should protect userspace app.
 		 * NOTE:  assumes 1:1 mapping here (that counters are organized
@@ -269,6 +363,7 @@ static int nmi_create_files(struct super_block *sb, struct dentry *root)
 		 */
 		if (unlikely(!avail_to_resrv_perfctr_nmi_bit(i)))
 			continue;
+#endif /* CONFIG_OPROFILE_EVENT_MULTIPLEX */
 
 		snprintf(buf,  sizeof(buf), "%d", i);
 		dir = oprofilefs_mkdir(sb, root, buf);
@@ -283,6 +378,57 @@ static int nmi_create_files(struct super_block *sb, struct dentry *root)
 	return 0;
 }
 
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void nmi_cpu_switch(void *dummy)
+{
+	int cpu = smp_processor_id();
+	int si = per_cpu(switch_index, cpu);
+	struct op_msrs *msrs = &per_cpu(cpu_msrs, cpu);
+
+	nmi_cpu_stop(NULL);
+	nmi_cpu_save_mpx_registers(msrs);
+
+	/* move to next set */
+	si += model->num_counters;
+	if ((si > model->num_virt_counters) || (counter_config[si].count == 0))
+		per_cpu(switch_index, cpu) = 0;
+	else
+		per_cpu(switch_index, cpu) = si;
+
+	model->switch_ctrl(model, msrs);
+	nmi_cpu_restore_mpx_registers(msrs);
+
+	nmi_cpu_start(NULL);
+}
+
+
+/*
+ * Quick check to see if multiplexing is necessary.
+ * The check should be sufficient since counters are used
+ * in ordre.
+ */
+static int nmi_multiplex_on(void)
+{
+	return counter_config[model->num_counters].count ? 0 : -EINVAL;
+}
+
+static int nmi_switch_event(void)
+{
+	if (!model->switch_ctrl)
+		return -ENOSYS;		/* not implemented */
+	if (nmi_multiplex_on() < 0)
+		return -EINVAL;		/* not necessary */
+
+	on_each_cpu(nmi_cpu_switch, NULL, 1);
+
+	atomic_inc(&multiplex_counter);
+
+	return 0;
+}
+
+#endif
+
 #ifdef CONFIG_SMP
 static int oprofile_cpu_notifier(struct notifier_block *b, unsigned long action,
 				 void *data)
@@ -516,12 +662,18 @@ int __init op_nmi_init(struct oprofile_operations *ops)
 	register_cpu_notifier(&oprofile_cpu_nb);
 #endif
 	/* default values, can be overwritten by model */
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	__raw_get_cpu_var(switch_index) = 0;
+#endif
 	ops->create_files	= nmi_create_files;
 	ops->setup		= nmi_setup;
 	ops->shutdown		= nmi_shutdown;
 	ops->start		= nmi_start;
 	ops->stop		= nmi_stop;
 	ops->cpu_type		= cpu_type;
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	ops->switch_events	= nmi_switch_event;
+#endif
 
 	if (model->init)
 		ret = model->init(ops);
diff --git a/arch/x86/oprofile/op_counter.h b/arch/x86/oprofile/op_counter.h
index 91b6a116165e..e28398df0df2 100644
--- a/arch/x86/oprofile/op_counter.h
+++ b/arch/x86/oprofile/op_counter.h
@@ -10,7 +10,7 @@
 #ifndef OP_COUNTER_H
 #define OP_COUNTER_H
 
-#define OP_MAX_COUNTER 8
+#define OP_MAX_COUNTER 32
 
 /* Per-perfctr configuration as set via
  * oprofilefs.
diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index f676f8825a3f..fdbed3a0c877 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -9,12 +9,15 @@
  * @author Philippe Elie
  * @author Graydon Hoare
  * @author Robert Richter <robert.richter@amd.com>
- * @author Barry Kasindorf
+ * @author Barry Kasindorf <barry.kasindorf@amd.com>
+ * @author Jason Yeh <jason.yeh@amd.com>
+ * @author Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
  */
 
 #include <linux/oprofile.h>
 #include <linux/device.h>
 #include <linux/pci.h>
+#include <linux/percpu.h>
 
 #include <asm/ptrace.h>
 #include <asm/msr.h>
@@ -25,12 +28,23 @@
 
 #define NUM_COUNTERS 4
 #define NUM_CONTROLS 4
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+#define NUM_VIRT_COUNTERS 32
+#define NUM_VIRT_CONTROLS 32
+#else
+#define NUM_VIRT_COUNTERS NUM_COUNTERS
+#define NUM_VIRT_CONTROLS NUM_CONTROLS
+#endif
+
 #define OP_EVENT_MASK			0x0FFF
 #define OP_CTR_OVERFLOW			(1ULL<<31)
 
 #define MSR_AMD_EVENTSEL_RESERVED	((0xFFFFFCF0ULL<<32)|(1ULL<<21))
 
-static unsigned long reset_value[NUM_COUNTERS];
+static unsigned long reset_value[NUM_VIRT_COUNTERS];
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+DECLARE_PER_CPU(int, switch_index);
+#endif
 
 #ifdef CONFIG_OPROFILE_IBS
 
@@ -82,6 +96,16 @@ static void op_amd_fill_in_addresses(struct op_msrs * const msrs)
 		else
 			msrs->controls[i].addr = 0;
 	}
+
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	for (i = 0; i < NUM_VIRT_COUNTERS; i++) {
+		int hw_counter = i % NUM_CONTROLS;
+		if (reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
+			msrs->multiplex[i].addr = MSR_K7_PERFCTR0 + hw_counter;
+		else
+			msrs->multiplex[i].addr = 0;
+	}
+#endif
 }
 
 static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
@@ -90,6 +114,15 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 	u64 val;
 	int i;
 
+	/* setup reset_value */
+	for (i = 0; i < NUM_VIRT_COUNTERS; ++i) {
+		if (counter_config[i].enabled) {
+			reset_value[i] = counter_config[i].count;
+		} else {
+			reset_value[i] = 0;
+		}
+	}
+
 	/* clear all counters */
 	for (i = 0; i < NUM_CONTROLS; ++i) {
 		if (unlikely(!msrs->controls[i].addr))
@@ -108,20 +141,49 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 
 	/* enable active counters */
 	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (counter_config[i].enabled && msrs->counters[i].addr) {
-			reset_value[i] = counter_config[i].count;
-			wrmsrl(msrs->counters[i].addr,
-			       -(u64)counter_config[i].count);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		int offset = i + __get_cpu_var(switch_index);
+#else
+		int offset = i;
+#endif
+		if (counter_config[offset].enabled && msrs->counters[i].addr) {
+			/* setup counter registers */
+			wrmsrl(msrs->counters[i].addr, -(u64)reset_value[offset]);
+
+			/* setup control registers */
 			rdmsrl(msrs->controls[i].addr, val);
 			val &= model->reserved;
-			val |= op_x86_get_ctrl(model, &counter_config[i]);
+			val |= op_x86_get_ctrl(model, &counter_config[offset]);
+			wrmsrl(msrs->controls[i].addr, val);
+		}
+	}
+}
+
+
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void op_amd_switch_ctrl(struct op_x86_model_spec const *model,
+			       struct op_msrs const * const msrs)
+{
+	u64 val;
+	int i;
+
+	/* enable active counters */
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		int offset = i + __get_cpu_var(switch_index);
+		if (counter_config[offset].enabled) {
+			/* setup control registers */
+			rdmsrl(msrs->controls[i].addr, val);
+			val &= model->reserved;
+			val |= op_x86_get_ctrl(model, &counter_config[offset]);
 			wrmsrl(msrs->controls[i].addr, val);
-		} else {
-			reset_value[i] = 0;
 		}
 	}
 }
 
+#endif
+
+
 #ifdef CONFIG_OPROFILE_IBS
 
 static inline int
@@ -230,14 +292,19 @@ static int op_amd_check_ctrs(struct pt_regs * const regs,
 	int i;
 
 	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (!reset_value[i])
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		int offset = i + __get_cpu_var(switch_index);
+#else
+		int offset = i;
+#endif
+		if (!reset_value[offset])
 			continue;
 		rdmsrl(msrs->counters[i].addr, val);
 		/* bit is clear if overflowed: */
 		if (val & OP_CTR_OVERFLOW)
 			continue;
-		oprofile_add_sample(regs, i);
-		wrmsrl(msrs->counters[i].addr, -(u64)reset_value[i]);
+		oprofile_add_sample(regs, offset);
+		wrmsrl(msrs->counters[i].addr, -(u64)reset_value[offset]);
 	}
 
 	op_amd_handle_ibs(regs, msrs);
@@ -250,8 +317,14 @@ static void op_amd_start(struct op_msrs const * const msrs)
 {
 	u64 val;
 	int i;
+
 	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (reset_value[i]) {
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		int offset = i + __get_cpu_var(switch_index);
+#else
+		int offset = i;
+#endif
+		if (reset_value[offset]) {
 			rdmsrl(msrs->controls[i].addr, val);
 			val |= ARCH_PERFMON_EVENTSEL0_ENABLE;
 			wrmsrl(msrs->controls[i].addr, val);
@@ -271,7 +344,11 @@ static void op_amd_stop(struct op_msrs const * const msrs)
 	 * pm callback
 	 */
 	for (i = 0; i < NUM_COUNTERS; ++i) {
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+		if (!reset_value[i + per_cpu(switch_index, smp_processor_id())])
+#else
 		if (!reset_value[i])
+#endif
 			continue;
 		rdmsrl(msrs->controls[i].addr, val);
 		val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE;
@@ -289,7 +366,7 @@ static void op_amd_shutdown(struct op_msrs const * const msrs)
 		if (msrs->counters[i].addr)
 			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
 	}
-	for (i = 0; i < NUM_CONTROLS; ++i) {
+	for (i = 0; i < NUM_COUNTERS; ++i) {
 		if (msrs->controls[i].addr)
 			release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
 	}
@@ -463,6 +540,8 @@ static void op_amd_exit(void) {}
 struct op_x86_model_spec const op_amd_spec = {
 	.num_counters		= NUM_COUNTERS,
 	.num_controls		= NUM_CONTROLS,
+	.num_virt_counters	= NUM_VIRT_COUNTERS,
+	.num_virt_controls	= NUM_VIRT_CONTROLS,
 	.reserved		= MSR_AMD_EVENTSEL_RESERVED,
 	.event_mask		= OP_EVENT_MASK,
 	.init			= op_amd_init,
@@ -473,4 +552,7 @@ struct op_x86_model_spec const op_amd_spec = {
 	.start			= &op_amd_start,
 	.stop			= &op_amd_stop,
 	.shutdown		= &op_amd_shutdown,
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	.switch_ctrl		= &op_amd_switch_ctrl,
+#endif
 };
diff --git a/arch/x86/oprofile/op_model_p4.c b/arch/x86/oprofile/op_model_p4.c
index 5921b7fc724b..65b9237cde8b 100644
--- a/arch/x86/oprofile/op_model_p4.c
+++ b/arch/x86/oprofile/op_model_p4.c
@@ -698,6 +698,8 @@ static void p4_shutdown(struct op_msrs const * const msrs)
 struct op_x86_model_spec const op_p4_ht2_spec = {
 	.num_counters		= NUM_COUNTERS_HT2,
 	.num_controls		= NUM_CONTROLS_HT2,
+	.num_virt_counters	= NUM_COUNTERS_HT2,
+	.num_virt_controls	= NUM_CONTROLS_HT2,
 	.fill_in_addresses	= &p4_fill_in_addresses,
 	.setup_ctrs		= &p4_setup_ctrs,
 	.check_ctrs		= &p4_check_ctrs,
@@ -710,6 +712,8 @@ struct op_x86_model_spec const op_p4_ht2_spec = {
 struct op_x86_model_spec const op_p4_spec = {
 	.num_counters		= NUM_COUNTERS_NON_HT,
 	.num_controls		= NUM_CONTROLS_NON_HT,
+	.num_virt_counters	= NUM_COUNTERS_NON_HT,
+	.num_virt_controls	= NUM_CONTROLS_NON_HT,
 	.fill_in_addresses	= &p4_fill_in_addresses,
 	.setup_ctrs		= &p4_setup_ctrs,
 	.check_ctrs		= &p4_check_ctrs,
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index 570d717c3308..098cbca5c0b0 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -206,6 +206,8 @@ static void ppro_shutdown(struct op_msrs const * const msrs)
 struct op_x86_model_spec const op_ppro_spec = {
 	.num_counters		= 2,
 	.num_controls		= 2,
+	.num_virt_counters	= 2,
+	.num_virt_controls	= 2,
 	.reserved		= MSR_PPRO_EVENTSEL_RESERVED,
 	.fill_in_addresses	= &ppro_fill_in_addresses,
 	.setup_ctrs		= &ppro_setup_ctrs,
diff --git a/arch/x86/oprofile/op_x86_model.h b/arch/x86/oprofile/op_x86_model.h
index 505489873b9d..0d07d23cb062 100644
--- a/arch/x86/oprofile/op_x86_model.h
+++ b/arch/x86/oprofile/op_x86_model.h
@@ -23,6 +23,7 @@ struct op_msr {
 struct op_msrs {
 	struct op_msr *counters;
 	struct op_msr *controls;
+	struct op_msr *multiplex;
 };
 
 struct pt_regs;
@@ -35,6 +36,8 @@ struct oprofile_operations;
 struct op_x86_model_spec {
 	unsigned int	num_counters;
 	unsigned int	num_controls;
+	unsigned int	num_virt_counters;
+	unsigned int	num_virt_controls;
 	u64		reserved;
 	u16		event_mask;
 	int		(*init)(struct oprofile_operations *ops);
@@ -47,6 +50,10 @@ struct op_x86_model_spec {
 	void		(*start)(struct op_msrs const * const msrs);
 	void		(*stop)(struct op_msrs const * const msrs);
 	void		(*shutdown)(struct op_msrs const * const msrs);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	void		(*switch_ctrl)(struct op_x86_model_spec const *model,
+				       struct op_msrs const * const msrs);
+#endif
 };
 
 struct op_counter_config;