1 files changed, 132 insertions, 72 deletions

diff --git a/arch/x86/events/amd/uncore.c b/arch/x86/events/amd/uncore.c
index a0b1bdb3ad42..4d1f7f2d9aff 100644
--- a/arch/x86/events/amd/uncore.c
+++ b/arch/x86/events/amd/uncore.c
@@ -22,13 +22,17 @@
 
 #define NUM_COUNTERS_NB		4
 #define NUM_COUNTERS_L2		4
-#define MAX_COUNTERS		NUM_COUNTERS_NB
+#define NUM_COUNTERS_L3		6
+#define MAX_COUNTERS		6
 
 #define RDPMC_BASE_NB		6
-#define RDPMC_BASE_L2		10
+#define RDPMC_BASE_LLC		10
 
 #define COUNTER_SHIFT		16
 
+static int num_counters_llc;
+static int num_counters_nb;
+
 static HLIST_HEAD(uncore_unused_list);
 
 struct amd_uncore {
@@ -45,30 +49,30 @@ struct amd_uncore {
 };
 
 static struct amd_uncore * __percpu *amd_uncore_nb;
-static struct amd_uncore * __percpu *amd_uncore_l2;
+static struct amd_uncore * __percpu *amd_uncore_llc;
 
 static struct pmu amd_nb_pmu;
-static struct pmu amd_l2_pmu;
+static struct pmu amd_llc_pmu;
 
 static cpumask_t amd_nb_active_mask;
-static cpumask_t amd_l2_active_mask;
+static cpumask_t amd_llc_active_mask;
 
 static bool is_nb_event(struct perf_event *event)
 {
 	return event->pmu->type == amd_nb_pmu.type;
 }
 
-static bool is_l2_event(struct perf_event *event)
+static bool is_llc_event(struct perf_event *event)
 {
-	return event->pmu->type == amd_l2_pmu.type;
+	return event->pmu->type == amd_llc_pmu.type;
 }
 
 static struct amd_uncore *event_to_amd_uncore(struct perf_event *event)
 {
 	if (is_nb_event(event) && amd_uncore_nb)
 		return *per_cpu_ptr(amd_uncore_nb, event->cpu);
-	else if (is_l2_event(event) && amd_uncore_l2)
-		return *per_cpu_ptr(amd_uncore_l2, event->cpu);
+	else if (is_llc_event(event) && amd_uncore_llc)
+		return *per_cpu_ptr(amd_uncore_llc, event->cpu);
 
 	return NULL;
 }
@@ -183,16 +187,16 @@ static int amd_uncore_event_init(struct perf_event *event)
 		return -ENOENT;
 
 	/*
-	 * NB and L2 counters (MSRs) are shared across all cores that share the
-	 * same NB / L2 cache. Interrupts can be directed to a single target
-	 * core, however, event counts generated by processes running on other
-	 * cores cannot be masked out. So we do not support sampling and
-	 * per-thread events.
+	 * NB and Last level cache counters (MSRs) are shared across all cores
+	 * that share the same NB / Last level cache. Interrupts can be directed
+	 * to a single target core, however, event counts generated by processes
+	 * running on other cores cannot be masked out. So we do not support
+	 * sampling and per-thread events.
 	 */
 	if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
 		return -EINVAL;
 
-	/* NB and L2 counters do not have usr/os/guest/host bits */
+	/* NB and Last level cache counters do not have usr/os/guest/host bits */
 	if (event->attr.exclude_user || event->attr.exclude_kernel ||
 	    event->attr.exclude_host || event->attr.exclude_guest)
 		return -EINVAL;
@@ -226,8 +230,8 @@ static ssize_t amd_uncore_attr_show_cpumask(struct device *dev,
 
 	if (pmu->type == amd_nb_pmu.type)
 		active_mask = &amd_nb_active_mask;
-	else if (pmu->type == amd_l2_pmu.type)
-		active_mask = &amd_l2_active_mask;
+	else if (pmu->type == amd_llc_pmu.type)
+		active_mask = &amd_llc_active_mask;
 	else
 		return 0;
 
@@ -244,30 +248,47 @@ static struct attribute_group amd_uncore_attr_group = {
 	.attrs = amd_uncore_attrs,
 };
 
-PMU_FORMAT_ATTR(event, "config:0-7,32-35");
-PMU_FORMAT_ATTR(umask, "config:8-15");
-
-static struct attribute *amd_uncore_format_attr[] = {
-	&format_attr_event.attr,
-	&format_attr_umask.attr,
-	NULL,
-};
-
-static struct attribute_group amd_uncore_format_group = {
-	.name = "format",
-	.attrs = amd_uncore_format_attr,
+/*
+ * Similar to PMU_FORMAT_ATTR but allowing for format_attr to be assigned based
+ * on family
+ */
+#define AMD_FORMAT_ATTR(_dev, _name, _format)				     \
+static ssize_t								     \
+_dev##_show##_name(struct device *dev,					     \
+		struct device_attribute *attr,				     \
+		char *page)						     \
+{									     \
+	BUILD_BUG_ON(sizeof(_format) >= PAGE_SIZE);			     \
+	return sprintf(page, _format "\n");				     \
+}									     \
+static struct device_attribute format_attr_##_dev##_name = __ATTR_RO(_dev);
+
+/* Used for each uncore counter type */
+#define AMD_ATTRIBUTE(_name)						     \
+static struct attribute *amd_uncore_format_attr_##_name[] = {		     \
+	&format_attr_event_##_name.attr,				     \
+	&format_attr_umask.attr,					     \
+	NULL,								     \
+};									     \
+static struct attribute_group amd_uncore_format_group_##_name = {	     \
+	.name = "format",						     \
+	.attrs = amd_uncore_format_attr_##_name,			     \
+};									     \
+static const struct attribute_group *amd_uncore_attr_groups_##_name[] = {    \
+	&amd_uncore_attr_group,						     \
+	&amd_uncore_format_group_##_name,				     \
+	NULL,								     \
 };
 
-static const struct attribute_group *amd_uncore_attr_groups[] = {
-	&amd_uncore_attr_group,
-	&amd_uncore_format_group,
-	NULL,
-};
+AMD_FORMAT_ATTR(event, , "config:0-7,32-35");
+AMD_FORMAT_ATTR(umask, , "config:8-15");
+AMD_FORMAT_ATTR(event, _df, "config:0-7,32-35,59-60");
+AMD_FORMAT_ATTR(event, _l3, "config:0-7");
+AMD_ATTRIBUTE(df);
+AMD_ATTRIBUTE(l3);
 
 static struct pmu amd_nb_pmu = {
 	.task_ctx_nr	= perf_invalid_context,
-	.attr_groups	= amd_uncore_attr_groups,
-	.name		= "amd_nb",
 	.event_init	= amd_uncore_event_init,
 	.add		= amd_uncore_add,
 	.del		= amd_uncore_del,
@@ -276,10 +297,8 @@ static struct pmu amd_nb_pmu = {
 	.read		= amd_uncore_read,
 };
 
-static struct pmu amd_l2_pmu = {
+static struct pmu amd_llc_pmu = {
 	.task_ctx_nr	= perf_invalid_context,
-	.attr_groups	= amd_uncore_attr_groups,
-	.name		= "amd_l2",
 	.event_init	= amd_uncore_event_init,
 	.add		= amd_uncore_add,
 	.del		= amd_uncore_del,
@@ -296,14 +315,14 @@ static struct amd_uncore *amd_uncore_alloc(unsigned int cpu)
 
 static int amd_uncore_cpu_up_prepare(unsigned int cpu)
 {
-	struct amd_uncore *uncore_nb = NULL, *uncore_l2;
+	struct amd_uncore *uncore_nb = NULL, *uncore_llc;
 
 	if (amd_uncore_nb) {
 		uncore_nb = amd_uncore_alloc(cpu);
 		if (!uncore_nb)
 			goto fail;
 		uncore_nb->cpu = cpu;
-		uncore_nb->num_counters = NUM_COUNTERS_NB;
+		uncore_nb->num_counters = num_counters_nb;
 		uncore_nb->rdpmc_base = RDPMC_BASE_NB;
 		uncore_nb->msr_base = MSR_F15H_NB_PERF_CTL;
 		uncore_nb->active_mask = &amd_nb_active_mask;
@@ -312,18 +331,18 @@ static int amd_uncore_cpu_up_prepare(unsigned int cpu)
 		*per_cpu_ptr(amd_uncore_nb, cpu) = uncore_nb;
 	}
 
-	if (amd_uncore_l2) {
-		uncore_l2 = amd_uncore_alloc(cpu);
-		if (!uncore_l2)
+	if (amd_uncore_llc) {
+		uncore_llc = amd_uncore_alloc(cpu);
+		if (!uncore_llc)
 			goto fail;
-		uncore_l2->cpu = cpu;
-		uncore_l2->num_counters = NUM_COUNTERS_L2;
-		uncore_l2->rdpmc_base = RDPMC_BASE_L2;
-		uncore_l2->msr_base = MSR_F16H_L2I_PERF_CTL;
-		uncore_l2->active_mask = &amd_l2_active_mask;
-		uncore_l2->pmu = &amd_l2_pmu;
-		uncore_l2->id = -1;
-		*per_cpu_ptr(amd_uncore_l2, cpu) = uncore_l2;
+		uncore_llc->cpu = cpu;
+		uncore_llc->num_counters = num_counters_llc;
+		uncore_llc->rdpmc_base = RDPMC_BASE_LLC;
+		uncore_llc->msr_base = MSR_F16H_L2I_PERF_CTL;
+		uncore_llc->active_mask = &amd_llc_active_mask;
+		uncore_llc->pmu = &amd_llc_pmu;
+		uncore_llc->id = -1;
+		*per_cpu_ptr(amd_uncore_llc, cpu) = uncore_llc;
 	}
 
 	return 0;
@@ -376,17 +395,17 @@ static int amd_uncore_cpu_starting(unsigned int cpu)
 		*per_cpu_ptr(amd_uncore_nb, cpu) = uncore;
 	}
 
-	if (amd_uncore_l2) {
+	if (amd_uncore_llc) {
 		unsigned int apicid = cpu_data(cpu).apicid;
 		unsigned int nshared;
 
-		uncore = *per_cpu_ptr(amd_uncore_l2, cpu);
+		uncore = *per_cpu_ptr(amd_uncore_llc, cpu);
 		cpuid_count(0x8000001d, 2, &eax, &ebx, &ecx, &edx);
 		nshared = ((eax >> 14) & 0xfff) + 1;
 		uncore->id = apicid - (apicid % nshared);
 
-		uncore = amd_uncore_find_online_sibling(uncore, amd_uncore_l2);
-		*per_cpu_ptr(amd_uncore_l2, cpu) = uncore;
+		uncore = amd_uncore_find_online_sibling(uncore, amd_uncore_llc);
+		*per_cpu_ptr(amd_uncore_llc, cpu) = uncore;
 	}
 
 	return 0;
@@ -419,8 +438,8 @@ static int amd_uncore_cpu_online(unsigned int cpu)
 	if (amd_uncore_nb)
 		uncore_online(cpu, amd_uncore_nb);
 
-	if (amd_uncore_l2)
-		uncore_online(cpu, amd_uncore_l2);
+	if (amd_uncore_llc)
+		uncore_online(cpu, amd_uncore_llc);
 
 	return 0;
 }
@@ -456,8 +475,8 @@ static int amd_uncore_cpu_down_prepare(unsigned int cpu)
 	if (amd_uncore_nb)
 		uncore_down_prepare(cpu, amd_uncore_nb);
 
-	if (amd_uncore_l2)
-		uncore_down_prepare(cpu, amd_uncore_l2);
+	if (amd_uncore_llc)
+		uncore_down_prepare(cpu, amd_uncore_llc);
 
 	return 0;
 }
@@ -479,8 +498,8 @@ static int amd_uncore_cpu_dead(unsigned int cpu)
 	if (amd_uncore_nb)
 		uncore_dead(cpu, amd_uncore_nb);
 
-	if (amd_uncore_l2)
-		uncore_dead(cpu, amd_uncore_l2);
+	if (amd_uncore_llc)
+		uncore_dead(cpu, amd_uncore_llc);
 
 	return 0;
 }
@@ -492,6 +511,47 @@ static int __init amd_uncore_init(void)
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
 		goto fail_nodev;
 
+	switch(boot_cpu_data.x86) {
+		case 23:
+			/* Family 17h: */
+			num_counters_nb = NUM_COUNTERS_NB;
+			num_counters_llc = NUM_COUNTERS_L3;
+			/*
+			 * For Family17h, the NorthBridge counters are
+			 * re-purposed as Data Fabric counters. Also, support is
+			 * added for L3 counters. The pmus are exported based on
+			 * family as either L2 or L3 and NB or DF.
+			 */
+			amd_nb_pmu.name = "amd_df";
+			amd_llc_pmu.name = "amd_l3";
+			format_attr_event_df.show = &event_show_df;
+			format_attr_event_l3.show = &event_show_l3;
+			break;
+		case 22:
+			/* Family 16h - may change: */
+			num_counters_nb = NUM_COUNTERS_NB;
+			num_counters_llc = NUM_COUNTERS_L2;
+			amd_nb_pmu.name = "amd_nb";
+			amd_llc_pmu.name = "amd_l2";
+			format_attr_event_df = format_attr_event;
+			format_attr_event_l3 = format_attr_event;
+			break;
+		default:
+			/*
+			 * All prior families have the same number of
+			 * NorthBridge and Last Level Cache counters
+			 */
+			num_counters_nb = NUM_COUNTERS_NB;
+			num_counters_llc = NUM_COUNTERS_L2;
+			amd_nb_pmu.name = "amd_nb";
+			amd_llc_pmu.name = "amd_l2";
+			format_attr_event_df = format_attr_event;
+			format_attr_event_l3 = format_attr_event;
+			break;
+	}
+	amd_nb_pmu.attr_groups = amd_uncore_attr_groups_df;
+	amd_llc_pmu.attr_groups = amd_uncore_attr_groups_l3;
+
 	if (!boot_cpu_has(X86_FEATURE_TOPOEXT))
 		goto fail_nodev;
 
@@ -510,16 +570,16 @@ static int __init amd_uncore_init(void)
 	}
 
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_L2)) {
-		amd_uncore_l2 = alloc_percpu(struct amd_uncore *);
-		if (!amd_uncore_l2) {
+		amd_uncore_llc = alloc_percpu(struct amd_uncore *);
+		if (!amd_uncore_llc) {
 			ret = -ENOMEM;
-			goto fail_l2;
+			goto fail_llc;
 		}
-		ret = perf_pmu_register(&amd_l2_pmu, amd_l2_pmu.name, -1);
+		ret = perf_pmu_register(&amd_llc_pmu, amd_llc_pmu.name, -1);
 		if (ret)
-			goto fail_l2;
+			goto fail_llc;
 
-		pr_info("perf: AMD L2I counters detected\n");
+		pr_info("perf: AMD LLC counters detected\n");
 		ret = 0;
 	}
 
@@ -529,7 +589,7 @@ static int __init amd_uncore_init(void)
 	if (cpuhp_setup_state(CPUHP_PERF_X86_AMD_UNCORE_PREP,
 			      "perf/x86/amd/uncore:prepare",
 			      amd_uncore_cpu_up_prepare, amd_uncore_cpu_dead))
-		goto fail_l2;
+		goto fail_llc;
 
 	if (cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING,
 			      "perf/x86/amd/uncore:starting",
@@ -546,11 +606,11 @@ fail_start:
 	cpuhp_remove_state(CPUHP_AP_PERF_X86_AMD_UNCORE_STARTING);
 fail_prep:
 	cpuhp_remove_state(CPUHP_PERF_X86_AMD_UNCORE_PREP);
-fail_l2:
+fail_llc:
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_NB))
 		perf_pmu_unregister(&amd_nb_pmu);
-	if (amd_uncore_l2)
-		free_percpu(amd_uncore_l2);
+	if (amd_uncore_llc)
+		free_percpu(amd_uncore_llc);
 fail_nb:
 	if (amd_uncore_nb)
 		free_percpu(amd_uncore_nb);