* [PATCH v2 1/6] perf/marvell: Marvell PEM performance monitor support
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 2/6] dt-bindings: perf: Add Marvell PEM performance monitor unit Gowthami Thiagarajan
` (4 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
PCI Express Interface PMU includes various performance counters to monitor
the data that is transmitted over the PCIe link. The counters track various
inbound and outbound transactions which includes separate counters for
posted/non-posted/completion TLPs. Also, inbound and outbound memory read
requests along with their latencies can also be monitored. Address
Translation Services(ATS)events such as ATS Translation, ATS Page Request,
ATS Invalidation along with their corresponding latencies are also
supported.
The performance counters are 64 bits wide.
For instance,
perf stat -e ib_tlp_pr <workload>
tracks the inbound posted TLPs for the workload.
Signed-off-by: Linu Cherian <lcherian@marvell.com>
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
- Added missed checks in event init
- Clear initialization of offset table with pem events
- Fixed comment style
MAINTAINERS | 6 +
drivers/perf/Kconfig | 7 +
drivers/perf/Makefile | 1 +
drivers/perf/marvell_pem_pmu.c | 437 +++++++++++++++++++++++++++++++++
include/linux/cpuhotplug.h | 1 +
5 files changed, 452 insertions(+)
create mode 100644 drivers/perf/marvell_pem_pmu.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 35e19594640d..9eaabaa60317 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -12578,6 +12578,12 @@ S: Supported
F: Documentation/networking/device_drivers/ethernet/marvell/octeontx2.rst
F: drivers/net/ethernet/marvell/octeontx2/af/
+MARVELL PEM PMU DRIVER
+M: Linu Cherian <lcherian@marvell.com>
+M: Gowthami Thiagarajan <gthiagarajan@marvell.com>
+S: Supported
+F: drivers/perf/marvell_pem_pmu.c
+
MARVELL PRESTERA ETHERNET SWITCH DRIVER
M: Taras Chornyi <taras.chornyi@plvision.eu>
S: Supported
diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index 711f82400086..b2da3ac067e3 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -213,4 +213,11 @@ source "drivers/perf/arm_cspmu/Kconfig"
source "drivers/perf/amlogic/Kconfig"
+config MARVELL_PEM_PMU
+ tristate "MARVELL PEM PMU Support"
+ depends on ARCH_THUNDER || (COMPILE_TEST && 64BIT)
+ help
+ Enable support for PCIe Interface performance monitoring
+ on Marvell platform.
+
endmenu
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index dabc859540ce..5ce7e5a8eeb1 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -20,6 +20,7 @@ obj-$(CONFIG_ARM_SPE_PMU) += arm_spe_pmu.o
obj-$(CONFIG_ARM_DMC620_PMU) += arm_dmc620_pmu.o
obj-$(CONFIG_MARVELL_CN10K_TAD_PMU) += marvell_cn10k_tad_pmu.o
obj-$(CONFIG_MARVELL_CN10K_DDR_PMU) += marvell_cn10k_ddr_pmu.o
+obj-$(CONFIG_MARVELL_PEM_PMU) += marvell_pem_pmu.o
obj-$(CONFIG_APPLE_M1_CPU_PMU) += apple_m1_cpu_pmu.o
obj-$(CONFIG_ALIBABA_UNCORE_DRW_PMU) += alibaba_uncore_drw_pmu.o
obj-$(CONFIG_ARM_CORESIGHT_PMU_ARCH_SYSTEM_PMU) += arm_cspmu/
diff --git a/drivers/perf/marvell_pem_pmu.c b/drivers/perf/marvell_pem_pmu.c
new file mode 100644
index 000000000000..ee6c4c80dd6d
--- /dev/null
+++ b/drivers/perf/marvell_pem_pmu.c
@@ -0,0 +1,437 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Marvell PEM(PCIe RC) Performance Monitor Driver
+ *
+ * Copyright (C) 2023 Marvell.
+ */
+
+#include <linux/acpi.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_device.h>
+#include <linux/perf_event.h>
+
+/*
+ * Each of these events maps to a free running 64 bit counter
+ * with no event control, but can be reset.
+ *
+ */
+enum pem_events {
+ IB_TLP_NPR,
+ IB_TLP_PR,
+ IB_TLP_CPL,
+ IB_TLP_DWORDS_NPR,
+ IB_TLP_DWORDS_PR,
+ IB_TLP_DWORDS_CPL,
+ IB_INFLIGHT,
+ IB_READS,
+ IB_REQ_NO_RO_NCB,
+ IB_REQ_NO_RO_EBUS,
+ OB_TLP_NPR,
+ OB_TLP_PR,
+ OB_TLP_CPL,
+ OB_TLP_DWORDS_NPR,
+ OB_TLP_DWORDS_PR,
+ OB_TLP_DWORDS_CPL,
+ OB_INFLIGHT,
+ OB_READS,
+ OB_MERGES_NPR,
+ OB_MERGES_PR,
+ OB_MERGES_CPL,
+ ATS_TRANS,
+ ATS_TRANS_LATENCY,
+ ATS_PRI,
+ ATS_PRI_LATENCY,
+ ATS_INV,
+ ATS_INV_LATENCY,
+ PEM_EVENTIDS_MAX,
+};
+
+static u64 eventid_to_offset_table[] = {
+ [IB_TLP_NPR] = 0x0,
+ [IB_TLP_PR] = 0x8,
+ [IB_TLP_CPL] = 0x10,
+ [IB_TLP_DWORDS_NPR] = 0x100,
+ [IB_TLP_DWORDS_PR] = 0x108,
+ [IB_TLP_DWORDS_CPL] = 0x110,
+ [IB_INFLIGHT] = 0x200,
+ [IB_READS] = 0x300,
+ [IB_REQ_NO_RO_NCB] = 0x400,
+ [IB_REQ_NO_RO_EBUS] = 0x408,
+ [OB_TLP_NPR] = 0x500,
+ [OB_TLP_PR] = 0x508,
+ [OB_TLP_CPL] = 0x510,
+ [OB_TLP_DWORDS_NPR] = 0x600,
+ [OB_TLP_DWORDS_PR] = 0x608,
+ [OB_TLP_DWORDS_CPL] = 0x610,
+ [OB_INFLIGHT] = 0x700,
+ [OB_READS] = 0x800,
+ [OB_MERGES_NPR] = 0x900,
+ [OB_MERGES_PR] = 0x908,
+ [OB_MERGES_CPL] = 0x910,
+ [ATS_TRANS] = 0x2D18,
+ [ATS_TRANS_LATENCY] = 0x2D20,
+ [ATS_PRI] = 0x2D28,
+ [ATS_PRI_LATENCY] = 0x2D30,
+ [ATS_INV] = 0x2D38,
+ [ATS_INV_LATENCY] = 0x2D40,
+};
+
+struct pem_pmu {
+ struct pmu pmu;
+ void __iomem *base;
+ unsigned int cpu;
+ struct device *dev;
+ struct hlist_node node;
+};
+
+#define to_pem_pmu(p) container_of(p, struct pem_pmu, pmu)
+
+static int eventid_to_offset(int eventid)
+{
+ return eventid_to_offset_table[eventid];
+}
+
+/* Events */
+static ssize_t pem_pmu_event_show(struct device *dev,
+ struct device_attribute *attr,
+ char *page)
+{
+ struct perf_pmu_events_attr *pmu_attr;
+
+ pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr);
+ return sysfs_emit(page, "event=0x%02llx\n", pmu_attr->id);
+}
+
+#define PEM_EVENT_ATTR(_name, _id) \
+ (&((struct perf_pmu_events_attr[]) { \
+ { .attr = __ATTR(_name, 0444, pem_pmu_event_show, NULL), \
+ .id = _id, } \
+ })[0].attr.attr)
+
+static struct attribute *pem_perf_events_attrs[] = {
+ PEM_EVENT_ATTR(ib_tlp_npr, IB_TLP_NPR),
+ PEM_EVENT_ATTR(ib_tlp_pr, IB_TLP_PR),
+ PEM_EVENT_ATTR(ib_tlp_cpl_partid, IB_TLP_CPL),
+ PEM_EVENT_ATTR(ib_tlp_dwords_npr, IB_TLP_DWORDS_NPR),
+ PEM_EVENT_ATTR(ib_tlp_dwords_pr, IB_TLP_DWORDS_PR),
+ PEM_EVENT_ATTR(ib_tlp_dwords_cpl_partid, IB_TLP_DWORDS_CPL),
+ PEM_EVENT_ATTR(ib_inflight, IB_INFLIGHT),
+ PEM_EVENT_ATTR(ib_reads, IB_READS),
+ PEM_EVENT_ATTR(ib_req_no_ro_ncb, IB_REQ_NO_RO_NCB),
+ PEM_EVENT_ATTR(ib_req_no_ro_ebus, IB_REQ_NO_RO_EBUS),
+ PEM_EVENT_ATTR(ob_tlp_npr_partid, OB_TLP_NPR),
+ PEM_EVENT_ATTR(ob_tlp_pr_partid, OB_TLP_PR),
+ PEM_EVENT_ATTR(ob_tlp_cpl_partid, OB_TLP_CPL),
+ PEM_EVENT_ATTR(ob_tlp_dwords_npr_partid, OB_TLP_DWORDS_NPR),
+ PEM_EVENT_ATTR(ob_tlp_dwords_pr_partid, OB_TLP_DWORDS_PR),
+ PEM_EVENT_ATTR(ob_tlp_dwords_cpl_partid, OB_TLP_DWORDS_CPL),
+ PEM_EVENT_ATTR(ob_inflight_partid, OB_INFLIGHT),
+ PEM_EVENT_ATTR(ob_reads_partid, OB_READS),
+ PEM_EVENT_ATTR(ob_merges_npr_partid, OB_MERGES_NPR),
+ PEM_EVENT_ATTR(ob_merges_pr_partid, OB_MERGES_PR),
+ PEM_EVENT_ATTR(ob_merges_cpl_partid, OB_MERGES_CPL),
+ PEM_EVENT_ATTR(ats_trans, ATS_TRANS),
+ PEM_EVENT_ATTR(ats_trans_latency, ATS_TRANS_LATENCY),
+ PEM_EVENT_ATTR(ats_pri, ATS_PRI),
+ PEM_EVENT_ATTR(ats_pri_latency, ATS_PRI_LATENCY),
+ PEM_EVENT_ATTR(ats_inv, ATS_INV),
+ PEM_EVENT_ATTR(ats_inv_latency, ATS_INV_LATENCY),
+ NULL
+};
+
+static struct attribute_group pem_perf_events_attr_group = {
+ .name = "events",
+ .attrs = pem_perf_events_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-5");
+
+static struct attribute *pem_perf_format_attrs[] = {
+ &format_attr_event.attr,
+ NULL
+};
+
+static struct attribute_group pem_perf_format_attr_group = {
+ .name = "format",
+ .attrs = pem_perf_format_attrs,
+};
+
+/* cpumask */
+static ssize_t pem_perf_cpumask_show(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct pem_pmu *pmu = dev_get_drvdata(dev);
+
+ return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
+}
+
+static struct device_attribute pem_perf_cpumask_attr =
+ __ATTR(cpumask, 0444, pem_perf_cpumask_show, NULL);
+
+static struct attribute *pem_perf_cpumask_attrs[] = {
+ &pem_perf_cpumask_attr.attr,
+ NULL
+};
+
+static struct attribute_group pem_perf_cpumask_attr_group = {
+ .attrs = pem_perf_cpumask_attrs,
+};
+
+static const struct attribute_group *pem_perf_attr_groups[] = {
+ &pem_perf_events_attr_group,
+ &pem_perf_cpumask_attr_group,
+ &pem_perf_format_attr_group,
+ NULL
+};
+
+static int pem_perf_event_init(struct perf_event *event)
+{
+ struct pem_pmu *pmu = to_pem_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+
+ if (event->attr.type != event->pmu->type)
+ return -ENOENT;
+
+ if (is_sampling_event(event) ||
+ event->attach_state & PERF_ATTACH_TASK) {
+ return -EOPNOTSUPP;
+ }
+
+ if (event->cpu < 0)
+ return -EOPNOTSUPP;
+
+ /* We must NOT create groups containing mixed PMUs */
+ if (event->group_leader->pmu != event->pmu &&
+ !is_software_event(event->group_leader))
+ return -EINVAL;
+
+ /*
+ * Set ownership of event to one CPU, same event can not be observed
+ * on multiple cpus at same time.
+ */
+ event->cpu = pmu->cpu;
+ hwc->idx = -1;
+ return 0;
+}
+
+static void pem_perf_counter_reset(struct pem_pmu *pmu,
+ struct perf_event *event, int eventid)
+{
+ writeq_relaxed(0x0, pmu->base + eventid_to_offset(eventid));
+}
+
+static u64 pem_perf_read_counter(struct pem_pmu *pmu,
+ struct perf_event *event, int eventid)
+{
+ return readq_relaxed(pmu->base + eventid_to_offset(eventid));
+}
+
+static void pem_perf_event_update(struct perf_event *event)
+{
+ struct pem_pmu *pmu = to_pem_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ u64 prev_count, new_count;
+
+ do {
+ prev_count = local64_read(&hwc->prev_count);
+ new_count = pem_perf_read_counter(pmu, event, hwc->idx);
+ } while (local64_xchg(&hwc->prev_count, new_count) != prev_count);
+
+ local64_add((new_count - prev_count), &event->count);
+}
+
+static void pem_perf_event_start(struct perf_event *event, int flags)
+{
+ struct pem_pmu *pmu = to_pem_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ int eventid = hwc->idx;
+
+ local64_set(&hwc->prev_count, 0);
+
+ pem_perf_counter_reset(pmu, event, eventid);
+
+ hwc->state = 0;
+}
+
+static int pem_perf_event_add(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ hwc->idx = event->attr.config;
+ if (hwc->idx >= PEM_EVENTIDS_MAX)
+ return -EINVAL;
+ hwc->state |= PERF_HES_STOPPED;
+
+ if (flags & PERF_EF_START)
+ pem_perf_event_start(event, flags);
+
+ return 0;
+}
+
+static void pem_perf_event_stop(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ if (flags & PERF_EF_UPDATE)
+ pem_perf_event_update(event);
+
+ hwc->state |= PERF_HES_STOPPED;
+}
+
+static void pem_perf_event_del(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ pem_perf_event_stop(event, PERF_EF_UPDATE);
+ hwc->idx = -1;
+}
+
+static int pem_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+ struct pem_pmu *pmu = hlist_entry_safe(node, struct pem_pmu,
+ node);
+ unsigned int target;
+
+ if (cpu != pmu->cpu)
+ return 0;
+
+ target = cpumask_any_but(cpu_online_mask, cpu);
+ if (target >= nr_cpu_ids)
+ return 0;
+
+ perf_pmu_migrate_context(&pmu->pmu, cpu, target);
+ pmu->cpu = target;
+ return 0;
+}
+
+static int pem_perf_probe(struct platform_device *pdev)
+{
+ struct pem_pmu *pem_pmu;
+ struct resource *res;
+ void __iomem *base;
+ char *name;
+ int ret;
+
+ pem_pmu = devm_kzalloc(&pdev->dev, sizeof(*pem_pmu), GFP_KERNEL);
+ if (!pem_pmu)
+ return -ENOMEM;
+
+ pem_pmu->dev = &pdev->dev;
+ platform_set_drvdata(pdev, pem_pmu);
+
+ base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+ if (IS_ERR(base))
+ return PTR_ERR(base);
+
+ pem_pmu->base = base;
+
+ pem_pmu->pmu = (struct pmu) {
+ .module = THIS_MODULE,
+ .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+ .task_ctx_nr = perf_invalid_context,
+ .attr_groups = pem_perf_attr_groups,
+ .event_init = pem_perf_event_init,
+ .add = pem_perf_event_add,
+ .del = pem_perf_event_del,
+ .start = pem_perf_event_start,
+ .stop = pem_perf_event_stop,
+ .read = pem_perf_event_update,
+ };
+
+ /* Choose this cpu to collect perf data */
+ pem_pmu->cpu = raw_smp_processor_id();
+
+ name = devm_kasprintf(pem_pmu->dev, GFP_KERNEL, "mrvl_pcie_rc_pmu_%llx",
+ res->start);
+ if (!name)
+ return -ENOMEM;
+
+ cpuhp_state_add_instance_nocalls
+ (CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE,
+ &pem_pmu->node);
+
+ ret = perf_pmu_register(&pem_pmu->pmu, name, -1);
+ if (ret)
+ goto error;
+
+ pr_info("Marvell PEM(PCIe RC) PMU Driver for pem@%llx\n", res->start);
+ return 0;
+error:
+ cpuhp_state_remove_instance_nocalls
+ (CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE,
+ &pem_pmu->node);
+ return ret;
+}
+
+static int pem_perf_remove(struct platform_device *pdev)
+{
+ struct pem_pmu *pem_pmu = platform_get_drvdata(pdev);
+
+ cpuhp_state_remove_instance_nocalls
+ (CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE,
+ &pem_pmu->node);
+
+ perf_pmu_unregister(&pem_pmu->pmu);
+ return 0;
+}
+
+#ifdef CONFIG_OF
+static const struct of_device_id pem_pmu_of_match[] = {
+ { .compatible = "marvell,odyssey-pem-pmu", },
+ { },
+};
+MODULE_DEVICE_TABLE(of, pem_pmu_of_match);
+#endif
+
+#ifdef CONFIG_ACPI
+static const struct acpi_device_id pem_pmu_acpi_match[] = {
+ {"MRVL000E", 0},
+ {},
+};
+MODULE_DEVICE_TABLE(acpi, pem_pmu_acpi_match);
+#endif
+
+static struct platform_driver pem_pmu_driver = {
+ .driver = {
+ .name = "pem-pmu",
+ .of_match_table = of_match_ptr(pem_pmu_of_match),
+ .acpi_match_table = ACPI_PTR(pem_pmu_acpi_match),
+ .suppress_bind_attrs = true,
+ },
+ .probe = pem_perf_probe,
+ .remove = pem_perf_remove,
+};
+
+static int __init pem_pmu_init(void)
+{
+ int ret;
+
+ ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE,
+ "perf/marvell/pem:online", NULL,
+ pem_pmu_offline_cpu);
+ if (ret)
+ return ret;
+
+ ret = platform_driver_register(&pem_pmu_driver);
+ if (ret)
+ cpuhp_remove_multi_state(CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE);
+ return ret;
+}
+
+static void __exit pem_pmu_exit(void)
+{
+ platform_driver_unregister(&pem_pmu_driver);
+ cpuhp_remove_multi_state(CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE);
+}
+
+module_init(pem_pmu_init);
+module_exit(pem_pmu_exit);
+
+MODULE_DESCRIPTION("Marvell PEM Perf driver");
+MODULE_AUTHOR("Linu Cherian <lcherian@marvell.com>");
+MODULE_LICENSE("GPL v2");
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index 3ceb9dfa0993..9649d5f4a0a6 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -236,6 +236,7 @@ enum cpuhp_state {
CPUHP_AP_PERF_ARM_APM_XGENE_ONLINE,
CPUHP_AP_PERF_ARM_CAVIUM_TX2_UNCORE_ONLINE,
CPUHP_AP_PERF_ARM_MARVELL_CN10K_DDR_ONLINE,
+ CPUHP_AP_PERF_ARM_MARVELL_PEM_ONLINE,
CPUHP_AP_PERF_POWERPC_NEST_IMC_ONLINE,
CPUHP_AP_PERF_POWERPC_CORE_IMC_ONLINE,
CPUHP_AP_PERF_POWERPC_THREAD_IMC_ONLINE,
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH v2 2/6] dt-bindings: perf: Add Marvell PEM performance monitor unit
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 1/6] perf/marvell: Marvell PEM " Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-14 8:13 ` Krzysztof Kozlowski
2023-08-12 14:25 ` [PATCH v2 3/6] perf/marvell : Odyssey LLC-TAD performance monitor support Gowthami Thiagarajan
` (3 subsequent siblings)
5 siblings, 1 reply; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
Add binding documentation for Marvell PEM performance monitor unit.
Signed-off-by: Linu Cherian <lcherian@marvell.com>
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
- Changed DT binding file name to match with compatible
- Added respective document in MAINTAINERS
.../perf/marvell,odyssey-pem-pmu.yaml | 38 +++++++++++++++++++
MAINTAINERS | 1 +
2 files changed, 39 insertions(+)
create mode 100644 Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
diff --git a/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml b/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
new file mode 100644
index 000000000000..a761fc09565b
--- /dev/null
+++ b/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
@@ -0,0 +1,38 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/perf/marvell,odyssey-pem-pmu.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Marvell PCIe interface performance monitor
+
+maintainers:
+ - Linu Cherian <lcherian@marvell.com>
+ - Gowthami Thiagarajan <gthiagarajan@marvell.com>
+
+properties:
+ compatible:
+ items:
+ - enum:
+ - marvell,odyssey-pem-pmu
+
+ reg:
+ maxItems: 1
+
+required:
+ - compatible
+ - reg
+
+additionalProperties: false
+
+examples:
+ - |
+ bus {
+ #address-cells = <2>;
+ #size-cells = <2>;
+
+ pmu@8e0000005000 {
+ compatible = "marvell,odyssey-pem-pmu";
+ reg = <0x8e00 0x00005000 0x0 0x3000>;
+ };
+ };
diff --git a/MAINTAINERS b/MAINTAINERS
index 9eaabaa60317..192a69d2fad6 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -12582,6 +12582,7 @@ MARVELL PEM PMU DRIVER
M: Linu Cherian <lcherian@marvell.com>
M: Gowthami Thiagarajan <gthiagarajan@marvell.com>
S: Supported
+F: Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
F: drivers/perf/marvell_pem_pmu.c
MARVELL PRESTERA ETHERNET SWITCH DRIVER
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH v2 2/6] dt-bindings: perf: Add Marvell PEM performance monitor unit
2023-08-12 14:25 ` [PATCH v2 2/6] dt-bindings: perf: Add Marvell PEM performance monitor unit Gowthami Thiagarajan
@ 2023-08-14 8:13 ` Krzysztof Kozlowski
0 siblings, 0 replies; 10+ messages in thread
From: Krzysztof Kozlowski @ 2023-08-14 8:13 UTC (permalink / raw)
To: Gowthami Thiagarajan, will, mark.rutland, linux-arm-kernel,
linux-kernel, robh+dt, krzysztof.kozlowski+dt, conor+dt,
devicetree
Cc: sgoutham, bbhushan2, lcherian
On 12/08/2023 16:25, Gowthami Thiagarajan wrote:
> Add binding documentation for Marvell PEM performance monitor unit.
>
> Signed-off-by: Linu Cherian <lcherian@marvell.com>
> Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
> ---
>
> v1->v2
> - Changed DT binding file name to match with compatible
> - Added respective document in MAINTAINERS
>
> .../perf/marvell,odyssey-pem-pmu.yaml | 38 +++++++++++++++++++
> MAINTAINERS | 1 +
> 2 files changed, 39 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
>
> diff --git a/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml b/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
> new file mode 100644
> index 000000000000..a761fc09565b
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
> @@ -0,0 +1,38 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/perf/marvell,odyssey-pem-pmu.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Marvell PCIe interface performance monitor
> +
> +maintainers:
> + - Linu Cherian <lcherian@marvell.com>
> + - Gowthami Thiagarajan <gthiagarajan@marvell.com>
> +
> +properties:
> + compatible:
> + items:
Drop items, you have just enum.
> + - enum:
> + - marvell,odyssey-pem-pmu
Neither commit msg nor description in the binding (there is none!) helps
me to understand what is the SoC here. Remember: compatibles are
specific to SoCs, not families.
Best regards,
Krzysztof
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v2 3/6] perf/marvell : Odyssey LLC-TAD performance monitor support
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 1/6] perf/marvell: Marvell PEM " Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 2/6] dt-bindings: perf: Add Marvell PEM performance monitor unit Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 4/6] dt-bindings: perf: Add Marvell Odyssey LLC-TAD pmu Gowthami Thiagarajan
` (2 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
Each TAD provides eight 64-bit counters for monitoring
cache behavior.The driver always configures the same counter for
all the TADs. The user would end up effectively reserving one of
eight counters in every TAD to look across all TADs.
The occurrences of events are aggregated and presented to the user
at the end of running the workload. The driver does not provide a
way for the user to partition TADs so that different TADs are used for
different applications.
The performance events reflect various internal or interface activities.
By combining the values from multiple performance counters, cache
performance can be measured in terms such as: cache miss rate, cache
allocations, interface retry rate, internal resource occupancy, etc.
Each supported counter's event and formatting information is exposed
to sysfs at /sys/devices/tad/. Use perf tool stat command to measure
the pmu events. For instance:
perf stat -e tad_hit_ltg,tad_hit_dtg <workload>
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
-Added missed checks in perf event init
-Fixed comment style
-Removed tad none event from the list
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
MAINTAINERS | 8 +
drivers/perf/Kconfig | 7 +
drivers/perf/Makefile | 1 +
drivers/perf/marvell_odyssey_tad_pmu.c | 424 +++++++++++++++++++++++++
4 files changed, 440 insertions(+)
create mode 100644 drivers/perf/marvell_odyssey_tad_pmu.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 192a69d2fad6..70d2971b93d4 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -12585,6 +12585,14 @@ S: Supported
F: Documentation/devicetree/bindings/perf/marvell,odyssey-pem-pmu.yaml
F: drivers/perf/marvell_pem_pmu.c
+MARVELL ODYSSEY TAD PMU DRIVER
+M: Gowthami Thiagarajan <gthiagarajan@marvell.com>
+M: Bharat Bhushan <bbhushan2@marvell.com>
+M: Linu Cherian <lcherian@marvell.com>
+M: George Cherian <gcherian@marvell.com>
+S: Supported
+F: drivers/perf/marvell_odyssey_tad_pmu.c
+
MARVELL PRESTERA ETHERNET SWITCH DRIVER
M: Taras Chornyi <taras.chornyi@plvision.eu>
S: Supported
diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index b2da3ac067e3..6fbfb5444f7c 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -220,4 +220,11 @@ config MARVELL_PEM_PMU
Enable support for PCIe Interface performance monitoring
on Marvell platform.
+config MARVELL_ODYSSEY_TAD_PMU
+ tristate "MARVELL ODYSSEY LLC-TAD PMU"
+ depends on ARCH_THUNDER || (COMPILE_TEST && 64BIT)
+ help
+ Provides support for Last-Level cache Tag-and-data Units (LLC-TAD)
+ performance monitor on Odyssey platform
+
endmenu
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index 5ce7e5a8eeb1..c2bc6123cca4 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -21,6 +21,7 @@ obj-$(CONFIG_ARM_DMC620_PMU) += arm_dmc620_pmu.o
obj-$(CONFIG_MARVELL_CN10K_TAD_PMU) += marvell_cn10k_tad_pmu.o
obj-$(CONFIG_MARVELL_CN10K_DDR_PMU) += marvell_cn10k_ddr_pmu.o
obj-$(CONFIG_MARVELL_PEM_PMU) += marvell_pem_pmu.o
+obj-$(CONFIG_MARVELL_ODYSSEY_TAD_PMU) += marvell_odyssey_tad_pmu.o
obj-$(CONFIG_APPLE_M1_CPU_PMU) += apple_m1_cpu_pmu.o
obj-$(CONFIG_ALIBABA_UNCORE_DRW_PMU) += alibaba_uncore_drw_pmu.o
obj-$(CONFIG_ARM_CORESIGHT_PMU_ARCH_SYSTEM_PMU) += arm_cspmu/
diff --git a/drivers/perf/marvell_odyssey_tad_pmu.c b/drivers/perf/marvell_odyssey_tad_pmu.c
new file mode 100644
index 000000000000..5e675b342598
--- /dev/null
+++ b/drivers/perf/marvell_odyssey_tad_pmu.c
@@ -0,0 +1,424 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Marvell Odyssey LLC-TAD perf driver
+ *
+ * Copyright (C) 2023 Marvell.
+ */
+
+#define pr_fmt(fmt) "tad_pmu: " fmt
+
+#include <linux/acpi.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_device.h>
+#include <linux/cpuhotplug.h>
+#include <linux/perf_event.h>
+#include <linux/platform_device.h>
+
+#define TAD_PFC_OFFSET 0x0800
+#define TAD_PFC(counter) (TAD_PFC_OFFSET | ((counter) << 3))
+#define TAD_PRF_OFFSET 0x0900
+#define TAD_PRF(counter) (TAD_PRF_OFFSET | ((counter) << 3))
+#define TAD_PRF_CNTSEL_MASK 0xFF
+#define TAD_MAX_COUNTERS 8
+
+#define to_tad_pmu(p) (container_of(p, struct tad_pmu, pmu))
+
+struct tad_region {
+ void __iomem *base;
+};
+
+struct tad_pmu {
+ struct pmu pmu;
+ struct tad_region *regions;
+ u32 region_cnt;
+ unsigned int cpu;
+ struct hlist_node node;
+ struct perf_event *events[TAD_MAX_COUNTERS];
+ DECLARE_BITMAP(counters_map, TAD_MAX_COUNTERS);
+};
+
+static int tad_pmu_cpuhp_state;
+
+static void tad_pmu_event_counter_read(struct perf_event *event)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ u32 counter_idx = hwc->idx;
+ u64 delta, prev, new;
+ int tad_region;
+
+ do {
+ prev = local64_read(&hwc->prev_count);
+ for (tad_region = 0, new = 0; tad_region < tad_pmu->region_cnt;
+ tad_region++)
+ new += readq(tad_pmu->regions[tad_region].base +
+ TAD_PFC(counter_idx));
+ } while (local64_cmpxchg(&hwc->prev_count, prev, new) != prev);
+
+ delta = (new - prev) & GENMASK_ULL(63, 0);
+ local64_add(delta, &event->count);
+}
+
+static void tad_pmu_event_counter_stop(struct perf_event *event, int flags)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ u32 counter_idx = hwc->idx;
+ int tad_region;
+
+ /*
+ * TAD()_PFC() stop counting on the write
+ * which sets TAD()_PRF()[CNTSEL] == 0
+ */
+ for (tad_region = 0; tad_region < tad_pmu->region_cnt;
+ tad_region++)
+ writeq(0, tad_pmu->regions[tad_region].base +
+ TAD_PRF(counter_idx));
+
+ tad_pmu_event_counter_read(event);
+ hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
+}
+
+static void tad_pmu_event_counter_start(struct perf_event *event, int flags)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ u32 event_idx = event->attr.config;
+ u32 counter_idx = hwc->idx;
+ u64 reg_val;
+ int tad_region;
+
+ hwc->state = 0;
+
+ /* Typically TAD_PFC() are zeroed to start counting */
+ for (tad_region = 0; tad_region < tad_pmu->region_cnt;
+ tad_region++)
+ writeq(0, tad_pmu->regions[tad_region].base +
+ TAD_PFC(counter_idx));
+
+ /*
+ * TAD()_PFC() start counting on the write
+ * which sets TAD()_PRF()[CNTSEL] != 0
+ */
+ for (tad_region = 0; tad_region < tad_pmu->region_cnt; tad_region++) {
+ reg_val = (event_idx & 0xFF);
+ writeq(reg_val, tad_pmu->regions[tad_region].base +
+ TAD_PRF(counter_idx));
+ }
+}
+
+static void tad_pmu_event_counter_del(struct perf_event *event, int flags)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ int idx = hwc->idx;
+
+ tad_pmu_event_counter_stop(event, flags | PERF_EF_UPDATE);
+ tad_pmu->events[idx] = NULL;
+ clear_bit(idx, tad_pmu->counters_map);
+}
+
+static int tad_pmu_event_counter_add(struct perf_event *event, int flags)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ int idx;
+
+ /* Get a free counter for this event */
+ idx = find_first_zero_bit(tad_pmu->counters_map, TAD_MAX_COUNTERS);
+ if (idx == TAD_MAX_COUNTERS)
+ return -EAGAIN;
+
+ set_bit(idx, tad_pmu->counters_map);
+
+ hwc->idx = idx;
+ hwc->state = PERF_HES_STOPPED;
+ tad_pmu->events[idx] = event;
+
+ if (flags & PERF_EF_START)
+ tad_pmu_event_counter_start(event, flags);
+
+ return 0;
+}
+
+static int tad_pmu_event_init(struct perf_event *event)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(event->pmu);
+
+ if (event->attr.type != event->pmu->type)
+ return -ENOENT;
+
+ if (is_sampling_event(event) ||
+ event->attach_state & PERF_ATTACH_TASK) {
+ return -EOPNOTSUPP;
+ }
+
+ if (!event->attr.disabled)
+ return -EINVAL;
+
+ if (event->state != PERF_EVENT_STATE_OFF)
+ return -EINVAL;
+
+ if (event->group_leader->pmu != event->pmu &&
+ !is_software_event(event->group_leader))
+ return -EINVAL;
+
+ event->cpu = tad_pmu->cpu;
+ event->hw.idx = -1;
+ event->hw.config_base = event->attr.config;
+
+ return 0;
+}
+
+static ssize_t tad_pmu_event_show(struct device *dev,
+ struct device_attribute *attr, char *page)
+{
+ struct perf_pmu_events_attr *pmu_attr;
+
+ pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr);
+ return sysfs_emit(page, "event=0x%02llx\n", pmu_attr->id);
+}
+
+#define TAD_PMU_EVENT_ATTR(_name, _id) \
+ (&((struct perf_pmu_events_attr[]) { \
+ { .attr = __ATTR(_name, 0444, tad_pmu_event_show, NULL),\
+ .id = _id, } \
+ })[0].attr.attr)
+
+static struct attribute *tad_pmu_event_attrs[] = {
+ TAD_PMU_EVENT_ATTR(tad_req_msh_in_exlmn, 0x3),
+ TAD_PMU_EVENT_ATTR(tad_alloc_dtg, 0x1a),
+ TAD_PMU_EVENT_ATTR(tad_alloc_ltg, 0x1b),
+ TAD_PMU_EVENT_ATTR(tad_alloc_any, 0x1c),
+ TAD_PMU_EVENT_ATTR(tad_hit_dtg, 0x1d),
+ TAD_PMU_EVENT_ATTR(tad_hit_ltg, 0x1e),
+ TAD_PMU_EVENT_ATTR(tad_hit_any, 0x1f),
+ TAD_PMU_EVENT_ATTR(tad_tag_rd, 0x20),
+ TAD_PMU_EVENT_ATTR(tad_tot_cycle, 0xFF),
+ NULL
+};
+
+static const struct attribute_group tad_pmu_events_attr_group = {
+ .name = "events",
+ .attrs = tad_pmu_event_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-7");
+
+static struct attribute *tad_pmu_format_attrs[] = {
+ &format_attr_event.attr,
+ NULL,
+};
+
+static struct attribute_group tad_pmu_format_attr_group = {
+ .name = "format",
+ .attrs = tad_pmu_format_attrs,
+};
+
+static ssize_t tad_pmu_cpumask_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct tad_pmu *tad_pmu = to_tad_pmu(dev_get_drvdata(dev));
+
+ return cpumap_print_to_pagebuf(true, buf, cpumask_of(tad_pmu->cpu));
+}
+
+static DEVICE_ATTR(cpumask, 0444, tad_pmu_cpumask_show, NULL);
+
+static struct attribute *tad_pmu_cpumask_attrs[] = {
+ &dev_attr_cpumask.attr,
+ NULL
+};
+
+static struct attribute_group tad_pmu_cpumask_attr_group = {
+ .attrs = tad_pmu_cpumask_attrs,
+};
+
+static const struct attribute_group *tad_pmu_attr_groups[] = {
+ &tad_pmu_events_attr_group,
+ &tad_pmu_format_attr_group,
+ &tad_pmu_cpumask_attr_group,
+ NULL
+};
+
+static int tad_pmu_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct tad_region *regions;
+ struct tad_pmu *tad_pmu;
+ struct resource *res;
+ u32 tad_pmu_page_size;
+ u32 tad_page_size;
+ u32 tad_cnt;
+ int i, ret;
+ char *name;
+
+ tad_pmu = devm_kzalloc(&pdev->dev, sizeof(*tad_pmu), GFP_KERNEL);
+ if (!tad_pmu)
+ return -ENOMEM;
+
+ platform_set_drvdata(pdev, tad_pmu);
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ if (!res) {
+ dev_err(&pdev->dev, "Mem resource not found\n");
+ return -ENODEV;
+ }
+
+ ret = device_property_read_u32(dev, "marvell,tad-page-size",
+ &tad_page_size);
+ if (ret) {
+ dev_err(&pdev->dev, "Can't find tad-page-size property\n");
+ return ret;
+ }
+
+ ret = device_property_read_u32(dev, "marvell,tad-pmu-page-size",
+ &tad_pmu_page_size);
+ if (ret) {
+ dev_err(&pdev->dev, "Can't find tad-pmu-page-size property\n");
+ return ret;
+ }
+
+ ret = device_property_read_u32(dev, "marvell,tad-cnt", &tad_cnt);
+ if (ret) {
+ dev_err(&pdev->dev, "Can't find tad-cnt property\n");
+ return ret;
+ }
+
+ regions = kcalloc(tad_cnt, sizeof(*regions), GFP_KERNEL);
+ if (!regions)
+ return -ENOMEM;
+
+ /* ioremap the distributed TAD pmu regions */
+ for (i = 0; i < tad_cnt && res->start < res->end; i++) {
+ regions[i].base = devm_ioremap(&pdev->dev,
+ res->start,
+ tad_pmu_page_size);
+ if (IS_ERR(regions[i].base)) {
+ dev_err(&pdev->dev, "TAD%d ioremap fail\n", i);
+ return -ENOMEM;
+ }
+ res->start += tad_page_size;
+ }
+
+ tad_pmu->regions = regions;
+ tad_pmu->region_cnt = tad_cnt;
+
+ tad_pmu->pmu = (struct pmu) {
+ .module = THIS_MODULE,
+ .attr_groups = tad_pmu_attr_groups,
+ .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+ .task_ctx_nr = perf_invalid_context,
+
+ .event_init = tad_pmu_event_init,
+ .add = tad_pmu_event_counter_add,
+ .del = tad_pmu_event_counter_del,
+ .start = tad_pmu_event_counter_start,
+ .stop = tad_pmu_event_counter_stop,
+ .read = tad_pmu_event_counter_read,
+ };
+
+ tad_pmu->cpu = raw_smp_processor_id();
+
+ /* Register pmu instance for cpu hotplug */
+ ret = cpuhp_state_add_instance_nocalls(tad_pmu_cpuhp_state,
+ &tad_pmu->node);
+ if (ret) {
+ dev_err(&pdev->dev, "Error %d registering hotplug\n", ret);
+ return ret;
+ }
+
+ name = "tad";
+ ret = perf_pmu_register(&tad_pmu->pmu, name, -1);
+ if (ret)
+ cpuhp_state_remove_instance_nocalls(tad_pmu_cpuhp_state,
+ &tad_pmu->node);
+
+ return ret;
+}
+
+static int tad_pmu_remove(struct platform_device *pdev)
+{
+ struct tad_pmu *pmu = platform_get_drvdata(pdev);
+
+ cpuhp_state_remove_instance_nocalls(tad_pmu_cpuhp_state,
+ &pmu->node);
+ perf_pmu_unregister(&pmu->pmu);
+
+ return 0;
+}
+
+#ifdef CONFIG_OF
+static const struct of_device_id tad_pmu_of_match[] = {
+ { .compatible = "marvell,odyssey-tad-pmu", },
+ {},
+};
+MODULE_DEVICE_TABLE(of, tad_pmu_of_match);
+#endif
+
+#ifdef CONFIG_ACPI
+static const struct acpi_device_id tad_pmu_acpi_match[] = {
+ {"MRVL000D", 0},
+ {},
+};
+MODULE_DEVICE_TABLE(acpi, tad_pmu_acpi_match);
+#endif
+
+static struct platform_driver odyssey_tad_pmu_driver = {
+ .driver = {
+ .name = "odyssey_tad_pmu",
+ .of_match_table = of_match_ptr(tad_pmu_of_match),
+ .acpi_match_table = ACPI_PTR(tad_pmu_acpi_match),
+ .suppress_bind_attrs = true,
+ },
+ .probe = tad_pmu_probe,
+ .remove = tad_pmu_remove,
+};
+
+static int odyssey_tad_pmu_offline_cpu(unsigned int cpu,
+ struct hlist_node *node)
+{
+ struct tad_pmu *pmu = hlist_entry_safe(node, struct tad_pmu, node);
+ unsigned int target;
+
+ if (cpu != pmu->cpu)
+ return 0;
+
+ target = cpumask_any_but(cpu_online_mask, cpu);
+ if (target >= nr_cpu_ids)
+ return 0;
+
+ perf_pmu_migrate_context(&pmu->pmu, cpu, target);
+ pmu->cpu = target;
+
+ return 0;
+}
+
+static int __init odyssey_tad_pmu_init(void)
+{
+ int ret;
+
+ ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
+ "perf/odyssey/tadpmu:online",
+ NULL,
+ odyssey_tad_pmu_offline_cpu);
+ if (ret < 0)
+ return ret;
+ tad_pmu_cpuhp_state = ret;
+ return platform_driver_register(&odyssey_tad_pmu_driver);
+}
+
+static void __exit odyssey_tad_pmu_exit(void)
+{
+ platform_driver_unregister(&odyssey_tad_pmu_driver);
+ cpuhp_remove_multi_state(tad_pmu_cpuhp_state);
+}
+
+module_init(odyssey_tad_pmu_init);
+module_exit(odyssey_tad_pmu_exit);
+
+MODULE_DESCRIPTION("Marvell ODYSSEY LLC-TAD Perf driver");
+MODULE_AUTHOR("Gowthami Thiagarajan <gthiagarajan@marvell.com>");
+MODULE_LICENSE("GPL v2");
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH v2 4/6] dt-bindings: perf: Add Marvell Odyssey LLC-TAD pmu
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
` (2 preceding siblings ...)
2023-08-12 14:25 ` [PATCH v2 3/6] perf/marvell : Odyssey LLC-TAD performance monitor support Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-14 8:15 ` Krzysztof Kozlowski
2023-08-12 14:25 ` [PATCH v2 5/6] perf/marvell: Odyssey DDR Performance monitor support Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 6/6] dt-bindings: perf: Add Marvell Odyssey DDR PMU Gowthami Thiagarajan
5 siblings, 1 reply; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
Add binding documentation for Marvell Odyssey LLC-TAD performance
monitor unit
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
- Changed DT binding file name to match with compatible
- Added respective document in MAINTAINERS
.../perf/marvell,odyssey-tad-pmu.yaml | 63 +++++++++++++++++++
MAINTAINERS | 1 +
2 files changed, 64 insertions(+)
create mode 100644 Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
diff --git a/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml b/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
new file mode 100644
index 000000000000..a1b9ee71e5f7
--- /dev/null
+++ b/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
@@ -0,0 +1,63 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/perf/marvell,odyssey-tad-pmu.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Marvell Odyssey LLC-TAD performance monitor
+
+maintainers:
+ - Gowthami Thiagarajan <gthiagarajan@marvell.com>
+
+description: |
+ The Tag-and-Data units (TADs) maintain coherence and contain CN10K
+ shared on-chip last level cache (LLC). The tad pmu measures the
+ performance of last-level cache. Each tad pmu supports up to eight
+ counters.
+
+ The DT setup comprises of number of tad blocks, the sizes of pmu
+ regions, tad blocks and overall base address of the HW.
+
+properties:
+ compatible:
+ const: marvell,odyssey-tad-pmu
+
+ reg:
+ maxItems: 1
+
+ marvell,tad-cnt:
+ description: specifies the number of tads on the soc
+ $ref: /schemas/types.yaml#/definitions/uint32
+
+ marvell,tad-page-size:
+ description: specifies the size of each tad page
+ $ref: /schemas/types.yaml#/definitions/uint32
+
+ marvell,tad-pmu-page-size:
+ description: specifies the size of page that the pmu uses
+ $ref: /schemas/types.yaml#/definitions/uint32
+
+required:
+ - compatible
+ - reg
+ - marvell,tad-cnt
+ - marvell,tad-page-size
+ - marvell,tad-pmu-page-size
+
+additionalProperties: false
+
+examples:
+ - |
+
+ tad {
+ #address-cells = <2>;
+ #size-cells = <2>;
+
+ pmu@87e22b030000 {
+ compatible = "marvell,odyssey-tad-pmu";
+ reg = <0x87e2 0x2b030000 0x0 0x1000>;
+ marvell,tad-cnt = <1>;
+ marvell,tad-page-size = <0x1000>;
+ marvell,tad-pmu-page-size = <0x1000>;
+ };
+ };
diff --git a/MAINTAINERS b/MAINTAINERS
index 70d2971b93d4..b9da1affe8b9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -12591,6 +12591,7 @@ M: Bharat Bhushan <bbhushan2@marvell.com>
M: Linu Cherian <lcherian@marvell.com>
M: George Cherian <gcherian@marvell.com>
S: Supported
+F: Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
F: drivers/perf/marvell_odyssey_tad_pmu.c
MARVELL PRESTERA ETHERNET SWITCH DRIVER
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH v2 4/6] dt-bindings: perf: Add Marvell Odyssey LLC-TAD pmu
2023-08-12 14:25 ` [PATCH v2 4/6] dt-bindings: perf: Add Marvell Odyssey LLC-TAD pmu Gowthami Thiagarajan
@ 2023-08-14 8:15 ` Krzysztof Kozlowski
0 siblings, 0 replies; 10+ messages in thread
From: Krzysztof Kozlowski @ 2023-08-14 8:15 UTC (permalink / raw)
To: Gowthami Thiagarajan, will, mark.rutland, linux-arm-kernel,
linux-kernel, robh+dt, krzysztof.kozlowski+dt, conor+dt,
devicetree
Cc: sgoutham, bbhushan2, lcherian
On 12/08/2023 16:25, Gowthami Thiagarajan wrote:
> Add binding documentation for Marvell Odyssey LLC-TAD performance
> monitor unit
>
> Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
> ---
>
> v1->v2
> - Changed DT binding file name to match with compatible
> - Added respective document in MAINTAINERS
>
> .../perf/marvell,odyssey-tad-pmu.yaml | 63 +++++++++++++++++++
> MAINTAINERS | 1 +
> 2 files changed, 64 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
>
> diff --git a/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml b/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
> new file mode 100644
> index 000000000000..a1b9ee71e5f7
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/perf/marvell,odyssey-tad-pmu.yaml
> @@ -0,0 +1,63 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/perf/marvell,odyssey-tad-pmu.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Marvell Odyssey LLC-TAD performance monitor
> +
> +maintainers:
> + - Gowthami Thiagarajan <gthiagarajan@marvell.com>
> +
> +description: |
> + The Tag-and-Data units (TADs) maintain coherence and contain CN10K
> + shared on-chip last level cache (LLC). The tad pmu measures the
> + performance of last-level cache. Each tad pmu supports up to eight
> + counters.
> +
> + The DT setup comprises of number of tad blocks, the sizes of pmu
> + regions, tad blocks and overall base address of the HW.
> +
> +properties:
> + compatible:
> + const: marvell,odyssey-tad-pmu
> +
> + reg:
> + maxItems: 1
> +
> + marvell,tad-cnt:
> + description: specifies the number of tads on the soc
Why would it differ? Each SoC is the same, right? Like factory makes 1
million of the same SoCs, so tad-cnt is the same on each of them
Please drop.
> + $ref: /schemas/types.yaml#/definitions/uint32
> +
> + marvell,tad-page-size:
> + description: specifies the size of each tad page
> + $ref: /schemas/types.yaml#/definitions/uint32
Same questions.
> +
> + marvell,tad-pmu-page-size:
> + description: specifies the size of page that the pmu uses
> + $ref: /schemas/types.yaml#/definitions/uint32
Same questions.
> +
> +required:
> + - compatible
> + - reg
> + - marvell,tad-cnt
> + - marvell,tad-page-size
> + - marvell,tad-pmu-page-size
> +
> +additionalProperties: false
> +
> +examples:
> + - |
> +
No need for blank line.
> + tad {
No clue what tad is, use some generic name.
> + #address-cells = <2>;
> + #size-cells = <2>;
> +
Best regards,
Krzysztof
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v2 5/6] perf/marvell: Odyssey DDR Performance monitor support
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
` (3 preceding siblings ...)
2023-08-12 14:25 ` [PATCH v2 4/6] dt-bindings: perf: Add Marvell Odyssey LLC-TAD pmu Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-12 14:25 ` [PATCH v2 6/6] dt-bindings: perf: Add Marvell Odyssey DDR PMU Gowthami Thiagarajan
5 siblings, 0 replies; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
Odyssey DRAM Subsystem supports eight counters for monitoring performance
and software can program those counters to monitor any of the defined
performance events. Supported performance events include those counted
at the interface between the DDR controller and the PHY, interface between
the DDR Controller and the CHI interconnect, or within the DDR Controller.
Additionally DSS also supports two fixed performance event counters, one
for ddr reads and the other for ddr writes.
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
-Fixed comment style and some indentations
drivers/perf/marvell_cn10k_ddr_pmu.c | 435 ++++++++++++++++++++++-----
1 file changed, 367 insertions(+), 68 deletions(-)
diff --git a/drivers/perf/marvell_cn10k_ddr_pmu.c b/drivers/perf/marvell_cn10k_ddr_pmu.c
index b94a5f6cc22b..3fed04bfab45 100644
--- a/drivers/perf/marvell_cn10k_ddr_pmu.c
+++ b/drivers/perf/marvell_cn10k_ddr_pmu.c
@@ -1,5 +1,6 @@
// SPDX-License-Identifier: GPL-2.0
-/* Marvell CN10K DRAM Subsystem (DSS) Performance Monitor Driver
+/*
+ * Marvell CN10K DRAM Subsystem (DSS) Performance Monitor Driver
*
* Copyright (C) 2021 Marvell.
*/
@@ -15,24 +16,29 @@
#include <linux/acpi.h>
/* Performance Counters Operating Mode Control Registers */
-#define DDRC_PERF_CNT_OP_MODE_CTRL 0x8020
-#define OP_MODE_CTRL_VAL_MANNUAL 0x1
+#define CN10K_DDRC_PERF_CNT_OP_MODE_CTRL 0x8020
+#define ODY_DDRC_PERF_CNT_OP_MODE_CTRL 0x20020
+#define OP_MODE_CTRL_VAL_MANUAL 0x1
/* Performance Counters Start Operation Control Registers */
-#define DDRC_PERF_CNT_START_OP_CTRL 0x8028
-#define START_OP_CTRL_VAL_START 0x1ULL
-#define START_OP_CTRL_VAL_ACTIVE 0x2
+#define CN10K_DDRC_PERF_CNT_START_OP_CTRL 0x8028
+#define ODY_DDRC_PERF_CNT_START_OP_CTRL 0x200A0
+#define START_OP_CTRL_VAL_START 0x1ULL
+#define START_OP_CTRL_VAL_ACTIVE 0x2
/* Performance Counters End Operation Control Registers */
-#define DDRC_PERF_CNT_END_OP_CTRL 0x8030
-#define END_OP_CTRL_VAL_END 0x1ULL
+#define CN10K_DDRC_PERF_CNT_END_OP_CTRL 0x8030
+#define ODY_DDRC_PERF_CNT_END_OP_CTRL 0x200E0
+#define END_OP_CTRL_VAL_END 0x1ULL
/* Performance Counters End Status Registers */
-#define DDRC_PERF_CNT_END_STATUS 0x8038
+#define CN10K_DDRC_PERF_CNT_END_STATUS 0x8038
+#define ODY_DDRC_PERF_CNT_END_STATUS 0x20120
#define END_STATUS_VAL_END_TIMER_MODE_END 0x1
/* Performance Counters Configuration Registers */
-#define DDRC_PERF_CFG_BASE 0x8040
+#define CN10K_DDRC_PERF_CFG_BASE 0x8040
+#define ODY_DDRC_PERF_CFG_BASE 0x20160
/* 8 Generic event counter + 2 fixed event counters */
#define DDRC_PERF_NUM_GEN_COUNTERS 8
@@ -43,18 +49,28 @@
DDRC_PERF_NUM_FIX_COUNTERS)
/* Generic event counter registers */
-#define DDRC_PERF_CFG(n) (DDRC_PERF_CFG_BASE + 8 * (n))
+#define DDRC_PERF_CFG(base, n) ((base) + 8 * (n))
#define EVENT_ENABLE BIT_ULL(63)
/* Two dedicated event counters for DDR reads and writes */
#define EVENT_DDR_READS 101
#define EVENT_DDR_WRITES 100
+#define DDRC_PERF_REG(base, n) ((base) + 8 * (n))
/*
* programmable events IDs in programmable event counters.
* DO NOT change these event-id numbers, they are used to
* program event bitmap in h/w.
+ *
*/
+/* Additional programmable events defined in Odyssey */
+#define EVENT_DFI_CMD_IS_RETRY 61
+#define EVENT_RD_UC_ECC_ERROR 60
+#define EVENT_RD_CRC_ERROR 59
+#define EVENT_CAPAR_ERROR 58
+#define EVENT_WR_CRC_ERROR 57
+#define EVENT_DFI_PARITY_POISON 56
+
#define EVENT_OP_IS_ZQLATCH 55
#define EVENT_OP_IS_ZQSTART 54
#define EVENT_OP_IS_TCR_MRR 53
@@ -64,8 +80,8 @@
#define EVENT_VISIBLE_WIN_LIMIT_REACHED_RD 49
#define EVENT_BSM_STARVATION 48
#define EVENT_BSM_ALLOC 47
-#define EVENT_LPR_REQ_WITH_NOCREDIT 46
-#define EVENT_HPR_REQ_WITH_NOCREDIT 45
+#define EVENT_RETRY_FIFO_FULL_OR_LPR_REQ_NOCRED 46
+#define EVENT_DFI_OR_HPR_REQ_NOCRED 45
#define EVENT_OP_IS_ZQCS 44
#define EVENT_OP_IS_ZQCL 43
#define EVENT_OP_IS_LOAD_MODE 42
@@ -103,28 +119,37 @@
#define EVENT_HIF_RD_OR_WR 1
/* Event counter value registers */
-#define DDRC_PERF_CNT_VALUE_BASE 0x8080
-#define DDRC_PERF_CNT_VALUE(n) (DDRC_PERF_CNT_VALUE_BASE + 8 * (n))
+#define CN10K_DDRC_PERF_CNT_VALUE_BASE 0x8080
+#define ODY_DDRC_PERF_CNT_VALUE_BASE 0x201C0
/* Fixed event counter enable/disable register */
-#define DDRC_PERF_CNT_FREERUN_EN 0x80C0
-#define DDRC_PERF_FREERUN_WRITE_EN 0x1
-#define DDRC_PERF_FREERUN_READ_EN 0x2
+#define CN10K_DDRC_PERF_CNT_FREERUN_EN 0x80C0
+#define DDRC_PERF_FREERUN_WRITE_EN 0x1
+#define DDRC_PERF_FREERUN_READ_EN 0x2
/* Fixed event counter control register */
-#define DDRC_PERF_CNT_FREERUN_CTRL 0x80C8
-#define DDRC_FREERUN_WRITE_CNT_CLR 0x1
-#define DDRC_FREERUN_READ_CNT_CLR 0x2
+#define CN10K_DDRC_PERF_CNT_FREERUN_CTRL 0x80C8
+#define ODY_DDRC_PERF_CNT_FREERUN_CTRL 0x20240
+#define DDRC_FREERUN_WRITE_CNT_CLR 0x1
+#define DDRC_FREERUN_READ_CNT_CLR 0x2
+
+/* Fixed event counter clear register, defined only for Odyssey */
+#define ODY_DDRC_PERF_CNT_FREERUN_CLR 0x20248
/* Fixed event counter value register */
-#define DDRC_PERF_CNT_VALUE_WR_OP 0x80D0
-#define DDRC_PERF_CNT_VALUE_RD_OP 0x80D8
-#define DDRC_PERF_CNT_VALUE_OVERFLOW BIT_ULL(48)
-#define DDRC_PERF_CNT_MAX_VALUE GENMASK_ULL(48, 0)
+#define CN10K_DDRC_PERF_CNT_VALUE_WR_OP 0x80D0
+#define CN10K_DDRC_PERF_CNT_VALUE_RD_OP 0x80D8
+#define ODY_DDRC_PERF_CNT_VALUE_WR_OP 0x20250
+#define ODY_DDRC_PERF_CNT_VALUE_RD_OP 0x20258
+
+#define VERSION_V1 1
+#define VERSION_V2 2
struct cn10k_ddr_pmu {
struct pmu pmu;
void __iomem *base;
+ struct ddr_pmu_platform_data *p_data;
+ int version;
unsigned int cpu;
struct device *dev;
int active_events;
@@ -135,6 +160,54 @@ struct cn10k_ddr_pmu {
#define to_cn10k_ddr_pmu(p) container_of(p, struct cn10k_ddr_pmu, pmu)
+struct ddr_pmu_platform_data {
+ u64 counter_overflow_val;
+ u64 counter_max_val;
+ u64 ddrc_perf_cnt_base;
+ u64 ddrc_perf_cfg_base;
+ u64 ddrc_perf_cnt_op_mode_ctrl;
+ u64 ddrc_perf_cnt_start_op_ctrl;
+ u64 ddrc_perf_cnt_end_op_ctrl;
+ u64 ddrc_perf_cnt_end_status;
+ u64 ddrc_perf_cnt_freerun_en;
+ u64 ddrc_perf_cnt_freerun_ctrl;
+ u64 ddrc_perf_cnt_freerun_clr;
+ u64 ddrc_perf_cnt_value_wr_op;
+ u64 ddrc_perf_cnt_value_rd_op;
+};
+
+static const struct ddr_pmu_platform_data cn10k_ddr_pmu_pdata = {
+ .counter_overflow_val = BIT_ULL(48),
+ .counter_max_val = GENMASK_ULL(48, 0),
+ .ddrc_perf_cnt_base = CN10K_DDRC_PERF_CNT_VALUE_BASE,
+ .ddrc_perf_cfg_base = CN10K_DDRC_PERF_CFG_BASE,
+ .ddrc_perf_cnt_op_mode_ctrl = CN10K_DDRC_PERF_CNT_OP_MODE_CTRL,
+ .ddrc_perf_cnt_start_op_ctrl = CN10K_DDRC_PERF_CNT_START_OP_CTRL,
+ .ddrc_perf_cnt_end_op_ctrl = CN10K_DDRC_PERF_CNT_END_OP_CTRL,
+ .ddrc_perf_cnt_end_status = CN10K_DDRC_PERF_CNT_END_STATUS,
+ .ddrc_perf_cnt_freerun_en = CN10K_DDRC_PERF_CNT_FREERUN_EN,
+ .ddrc_perf_cnt_freerun_ctrl = CN10K_DDRC_PERF_CNT_FREERUN_CTRL,
+ .ddrc_perf_cnt_freerun_clr = 0,
+ .ddrc_perf_cnt_value_wr_op = CN10K_DDRC_PERF_CNT_VALUE_WR_OP,
+ .ddrc_perf_cnt_value_rd_op = CN10K_DDRC_PERF_CNT_VALUE_RD_OP,
+};
+
+static const struct ddr_pmu_platform_data odyssey_ddr_pmu_pdata = {
+ .counter_overflow_val = 0,
+ .counter_max_val = GENMASK_ULL(63, 0),
+ .ddrc_perf_cnt_base = ODY_DDRC_PERF_CNT_VALUE_BASE,
+ .ddrc_perf_cfg_base = ODY_DDRC_PERF_CFG_BASE,
+ .ddrc_perf_cnt_op_mode_ctrl = ODY_DDRC_PERF_CNT_OP_MODE_CTRL,
+ .ddrc_perf_cnt_start_op_ctrl = ODY_DDRC_PERF_CNT_START_OP_CTRL,
+ .ddrc_perf_cnt_end_op_ctrl = ODY_DDRC_PERF_CNT_END_OP_CTRL,
+ .ddrc_perf_cnt_end_status = ODY_DDRC_PERF_CNT_END_STATUS,
+ .ddrc_perf_cnt_freerun_en = 0,
+ .ddrc_perf_cnt_freerun_ctrl = ODY_DDRC_PERF_CNT_FREERUN_CTRL,
+ .ddrc_perf_cnt_freerun_clr = ODY_DDRC_PERF_CNT_FREERUN_CLR,
+ .ddrc_perf_cnt_value_wr_op = ODY_DDRC_PERF_CNT_VALUE_WR_OP,
+ .ddrc_perf_cnt_value_rd_op = ODY_DDRC_PERF_CNT_VALUE_RD_OP,
+};
+
static ssize_t cn10k_ddr_pmu_event_show(struct device *dev,
struct device_attribute *attr,
char *page)
@@ -190,9 +263,9 @@ static struct attribute *cn10k_ddr_perf_events_attrs[] = {
CN10K_DDR_PMU_EVENT_ATTR(ddr_zqcl, EVENT_OP_IS_ZQCL),
CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_wr_access, EVENT_OP_IS_ZQCS),
CN10K_DDR_PMU_EVENT_ATTR(ddr_hpr_req_with_nocredit,
- EVENT_HPR_REQ_WITH_NOCREDIT),
+ EVENT_DFI_OR_HPR_REQ_NOCRED),
CN10K_DDR_PMU_EVENT_ATTR(ddr_lpr_req_with_nocredit,
- EVENT_LPR_REQ_WITH_NOCREDIT),
+ EVENT_RETRY_FIFO_FULL_OR_LPR_REQ_NOCRED),
CN10K_DDR_PMU_EVENT_ATTR(ddr_bsm_alloc, EVENT_BSM_ALLOC),
CN10K_DDR_PMU_EVENT_ATTR(ddr_bsm_starvation, EVENT_BSM_STARVATION),
CN10K_DDR_PMU_EVENT_ATTR(ddr_win_limit_reached_rd,
@@ -215,6 +288,85 @@ static struct attribute_group cn10k_ddr_perf_events_attr_group = {
.attrs = cn10k_ddr_perf_events_attrs,
};
+static struct attribute *odyssey_ddr_perf_events_attrs[] = {
+ /* Programmable */
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hif_rd_or_wr_access, EVENT_HIF_RD_OR_WR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hif_wr_access, EVENT_HIF_WR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hif_rd_access, EVENT_HIF_RD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hif_rmw_access, EVENT_HIF_RMW),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hif_pri_rdaccess, EVENT_HIF_HI_PRI_RD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_rd_bypass_access, EVENT_READ_BYPASS),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_act_bypass_access, EVENT_ACT_BYPASS),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dfi_wr_data_access,
+ EVENT_DFI_WR_DATA_CYCLES),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dfi_rd_data_access,
+ EVENT_DFI_RD_DATA_CYCLES),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_hpri_sched_rd_crit_access,
+ EVENT_HPR_XACT_WHEN_CRITICAL),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_lpri_sched_rd_crit_access,
+ EVENT_LPR_XACT_WHEN_CRITICAL),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_wr_trxn_crit_access,
+ EVENT_WR_XACT_WHEN_CRITICAL),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_active_access, EVENT_OP_IS_ACTIVATE),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_rd_or_wr_access,
+ EVENT_OP_IS_RD_OR_WR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_rd_active_access,
+ EVENT_OP_IS_RD_ACTIVATE),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_read, EVENT_OP_IS_RD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_write, EVENT_OP_IS_WR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_mwr, EVENT_OP_IS_MWR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_precharge, EVENT_OP_IS_PRECHARGE),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_precharge_for_rdwr,
+ EVENT_PRECHARGE_FOR_RDWR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_precharge_for_other,
+ EVENT_PRECHARGE_FOR_OTHER),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_rdwr_transitions, EVENT_RDWR_TRANSITIONS),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_write_combine, EVENT_WRITE_COMBINE),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_war_hazard, EVENT_WAR_HAZARD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_raw_hazard, EVENT_RAW_HAZARD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_waw_hazard, EVENT_WAW_HAZARD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_enter_selfref, EVENT_OP_IS_ENTER_SELFREF),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_enter_powerdown,
+ EVENT_OP_IS_ENTER_POWERDOWN),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_enter_mpsm, EVENT_OP_IS_ENTER_MPSM),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_refresh, EVENT_OP_IS_REFRESH),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_crit_ref, EVENT_OP_IS_CRIT_REF),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_spec_ref, EVENT_OP_IS_SPEC_REF),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_load_mode, EVENT_OP_IS_LOAD_MODE),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_zqcl, EVENT_OP_IS_ZQCL),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_cam_wr_access, EVENT_OP_IS_ZQCS),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dfi_cycles, EVENT_DFI_OR_HPR_REQ_NOCRED),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_retry_fifo_full,
+ EVENT_RETRY_FIFO_FULL_OR_LPR_REQ_NOCRED),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_bsm_alloc, EVENT_BSM_ALLOC),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_bsm_starvation, EVENT_BSM_STARVATION),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_win_limit_reached_rd,
+ EVENT_VISIBLE_WIN_LIMIT_REACHED_RD),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_win_limit_reached_wr,
+ EVENT_VISIBLE_WIN_LIMIT_REACHED_WR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dqsosc_mpc, EVENT_OP_IS_DQSOSC_MPC),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dqsosc_mrr, EVENT_OP_IS_DQSOSC_MRR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_tcr_mrr, EVENT_OP_IS_TCR_MRR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_zqstart, EVENT_OP_IS_ZQSTART),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_zqlatch, EVENT_OP_IS_ZQLATCH),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dfi_parity_poison,
+ EVENT_DFI_PARITY_POISON),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_wr_crc_error, EVENT_WR_CRC_ERROR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_capar_error, EVENT_CAPAR_ERROR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_rd_crc_error, EVENT_RD_CRC_ERROR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_rd_uc_ecc_error, EVENT_RD_UC_ECC_ERROR),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_dfi_cmd_is_retry, EVENT_DFI_CMD_IS_RETRY),
+ /* Free run event counters */
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_ddr_reads, EVENT_DDR_READS),
+ CN10K_DDR_PMU_EVENT_ATTR(ddr_ddr_writes, EVENT_DDR_WRITES),
+ NULL
+};
+
+static struct attribute_group odyssey_ddr_perf_events_attr_group = {
+ .name = "events",
+ .attrs = odyssey_ddr_perf_events_attrs,
+};
+
PMU_FORMAT_ATTR(event, "config:0-8");
static struct attribute *cn10k_ddr_perf_format_attrs[] = {
@@ -255,6 +407,13 @@ static const struct attribute_group *cn10k_attr_groups[] = {
NULL,
};
+static const struct attribute_group *odyssey_attr_groups[] = {
+ &odyssey_ddr_perf_events_attr_group,
+ &cn10k_ddr_perf_format_attr_group,
+ &cn10k_ddr_perf_cpumask_attr_group,
+ NULL
+};
+
/* Default poll timeout is 100 sec, which is very sufficient for
* 48 bit counter incremented max at 5.6 GT/s, which may take many
* hours to overflow.
@@ -267,13 +426,24 @@ static ktime_t cn10k_ddr_pmu_timer_period(void)
return ms_to_ktime((u64)cn10k_ddr_pmu_poll_period_sec * USEC_PER_SEC);
}
-static int ddr_perf_get_event_bitmap(int eventid, u64 *event_bitmap)
+static int ddr_perf_get_event_bitmap(int eventid, u64 *event_bitmap,
+ struct cn10k_ddr_pmu *ddr_pmu)
{
+ int ret = 0;
+
switch (eventid) {
case EVENT_HIF_RD_OR_WR ... EVENT_WAW_HAZARD:
case EVENT_OP_IS_REFRESH ... EVENT_OP_IS_ZQLATCH:
*event_bitmap = (1ULL << (eventid - 1));
break;
+ case EVENT_DFI_PARITY_POISON ...EVENT_DFI_CMD_IS_RETRY:
+ if (ddr_pmu->version == VERSION_V2) {
+ *event_bitmap = (1ULL << (eventid - 1));
+ } else {
+ pr_err("%s Invalid eventid %d\n", __func__, eventid);
+ ret = -EINVAL;
+ }
+ break;
case EVENT_OP_IS_ENTER_SELFREF:
case EVENT_OP_IS_ENTER_POWERDOWN:
case EVENT_OP_IS_ENTER_MPSM:
@@ -281,10 +451,10 @@ static int ddr_perf_get_event_bitmap(int eventid, u64 *event_bitmap)
break;
default:
pr_err("%s Invalid eventid %d\n", __func__, eventid);
- return -EINVAL;
+ ret = -EINVAL;
}
- return 0;
+ return ret;
}
static int cn10k_ddr_perf_alloc_counter(struct cn10k_ddr_pmu *pmu,
@@ -357,6 +527,7 @@ static void cn10k_ddr_perf_counter_enable(struct cn10k_ddr_pmu *pmu,
{
u32 reg;
u64 val;
+ struct ddr_pmu_platform_data *p_data = pmu->p_data;
if (counter > DDRC_PERF_NUM_COUNTERS) {
pr_err("Error: unsupported counter %d\n", counter);
@@ -364,7 +535,7 @@ static void cn10k_ddr_perf_counter_enable(struct cn10k_ddr_pmu *pmu,
}
if (counter < DDRC_PERF_NUM_GEN_COUNTERS) {
- reg = DDRC_PERF_CFG(counter);
+ reg = DDRC_PERF_CFG(p_data->ddrc_perf_cfg_base, counter);
val = readq_relaxed(pmu->base + reg);
if (enable)
@@ -374,7 +545,13 @@ static void cn10k_ddr_perf_counter_enable(struct cn10k_ddr_pmu *pmu,
writeq_relaxed(val, pmu->base + reg);
} else {
- val = readq_relaxed(pmu->base + DDRC_PERF_CNT_FREERUN_EN);
+ if (p_data->ddrc_perf_cnt_freerun_en)
+ val = readq_relaxed(pmu->base +
+ p_data->ddrc_perf_cnt_freerun_en);
+ else
+ val = readq_relaxed(pmu->base +
+ p_data->ddrc_perf_cnt_freerun_ctrl);
+
if (enable) {
if (counter == DDRC_PERF_READ_COUNTER_IDX)
val |= DDRC_PERF_FREERUN_READ_EN;
@@ -386,7 +563,13 @@ static void cn10k_ddr_perf_counter_enable(struct cn10k_ddr_pmu *pmu,
else
val &= ~DDRC_PERF_FREERUN_WRITE_EN;
}
- writeq_relaxed(val, pmu->base + DDRC_PERF_CNT_FREERUN_EN);
+
+ if (p_data->ddrc_perf_cnt_freerun_en)
+ writeq_relaxed(val, pmu->base +
+ p_data->ddrc_perf_cnt_freerun_en);
+ else
+ writeq_relaxed(val, pmu->base +
+ p_data->ddrc_perf_cnt_freerun_ctrl);
}
}
@@ -394,13 +577,18 @@ static u64 cn10k_ddr_perf_read_counter(struct cn10k_ddr_pmu *pmu, int counter)
{
u64 val;
+ struct ddr_pmu_platform_data *p_data = pmu->p_data;
+
if (counter == DDRC_PERF_READ_COUNTER_IDX)
- return readq_relaxed(pmu->base + DDRC_PERF_CNT_VALUE_RD_OP);
+ return readq_relaxed(pmu->base +
+ p_data->ddrc_perf_cnt_value_rd_op);
if (counter == DDRC_PERF_WRITE_COUNTER_IDX)
- return readq_relaxed(pmu->base + DDRC_PERF_CNT_VALUE_WR_OP);
+ return readq_relaxed(pmu->base +
+ p_data->ddrc_perf_cnt_value_wr_op);
- val = readq_relaxed(pmu->base + DDRC_PERF_CNT_VALUE(counter));
+ val = readq_relaxed(pmu->base +
+ DDRC_PERF_REG(p_data->ddrc_perf_cnt_base, counter));
return val;
}
@@ -408,6 +596,7 @@ static void cn10k_ddr_perf_event_update(struct perf_event *event)
{
struct cn10k_ddr_pmu *pmu = to_cn10k_ddr_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
+ struct ddr_pmu_platform_data *p_data = pmu->p_data;
u64 prev_count, new_count, mask;
do {
@@ -415,20 +604,48 @@ static void cn10k_ddr_perf_event_update(struct perf_event *event)
new_count = cn10k_ddr_perf_read_counter(pmu, hwc->idx);
} while (local64_xchg(&hwc->prev_count, new_count) != prev_count);
- mask = DDRC_PERF_CNT_MAX_VALUE;
+ mask = p_data->counter_max_val;
local64_add((new_count - prev_count) & mask, &event->count);
}
+static void cn10k_ddr_perf_counter_start(struct cn10k_ddr_pmu *ddr_pmu,
+ int counter)
+{
+ struct ddr_pmu_platform_data *p_data = ddr_pmu->p_data;
+ u64 ctrl_reg = p_data->ddrc_perf_cnt_start_op_ctrl;
+
+ writeq_relaxed(START_OP_CTRL_VAL_START, ddr_pmu->base +
+ DDRC_PERF_REG(ctrl_reg, counter));
+}
+
+static void cn10k_ddr_perf_counter_stop(struct cn10k_ddr_pmu *ddr_pmu,
+ int counter)
+{
+ struct ddr_pmu_platform_data *p_data = ddr_pmu->p_data;
+ u64 ctrl_reg = p_data->ddrc_perf_cnt_end_op_ctrl;
+
+ writeq_relaxed(END_OP_CTRL_VAL_END, ddr_pmu->base +
+ DDRC_PERF_REG(ctrl_reg, counter));
+}
+
static void cn10k_ddr_perf_event_start(struct perf_event *event, int flags)
{
struct cn10k_ddr_pmu *pmu = to_cn10k_ddr_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
int counter = hwc->idx;
+ u64 ctrl_reg = pmu->p_data->ddrc_perf_cnt_op_mode_ctrl;
local64_set(&hwc->prev_count, 0);
cn10k_ddr_perf_counter_enable(pmu, counter, true);
+ if (pmu->version == VERSION_V2) {
+ /* Setup the PMU counter to work in manual mode */
+ writeq_relaxed(OP_MODE_CTRL_VAL_MANUAL, pmu->base +
+ DDRC_PERF_REG(ctrl_reg, counter));
+
+ cn10k_ddr_perf_counter_start(pmu, counter);
+ }
hwc->state = 0;
}
@@ -436,6 +653,7 @@ static void cn10k_ddr_perf_event_start(struct perf_event *event, int flags)
static int cn10k_ddr_perf_event_add(struct perf_event *event, int flags)
{
struct cn10k_ddr_pmu *pmu = to_cn10k_ddr_pmu(event->pmu);
+ struct ddr_pmu_platform_data *p_data = pmu->p_data;
struct hw_perf_event *hwc = &event->hw;
u8 config = event->attr.config;
int counter, ret;
@@ -455,8 +673,8 @@ static int cn10k_ddr_perf_event_add(struct perf_event *event, int flags)
if (counter < DDRC_PERF_NUM_GEN_COUNTERS) {
/* Generic counters, configure event id */
- reg_offset = DDRC_PERF_CFG(counter);
- ret = ddr_perf_get_event_bitmap(config, &val);
+ reg_offset = DDRC_PERF_CFG(p_data->ddrc_perf_cfg_base, counter);
+ ret = ddr_perf_get_event_bitmap(config, &val, pmu);
if (ret)
return ret;
@@ -468,7 +686,12 @@ static int cn10k_ddr_perf_event_add(struct perf_event *event, int flags)
else
val = DDRC_FREERUN_WRITE_CNT_CLR;
- writeq_relaxed(val, pmu->base + DDRC_PERF_CNT_FREERUN_CTRL);
+ if (p_data->ddrc_perf_cnt_freerun_clr)
+ writeq_relaxed(val, pmu->base +
+ p_data->ddrc_perf_cnt_freerun_clr);
+ else
+ writeq_relaxed(val, pmu->base +
+ p_data->ddrc_perf_cnt_freerun_ctrl);
}
hwc->state |= PERF_HES_STOPPED;
@@ -487,6 +710,9 @@ static void cn10k_ddr_perf_event_stop(struct perf_event *event, int flags)
cn10k_ddr_perf_counter_enable(pmu, counter, false);
+ if (pmu->version == VERSION_V2)
+ cn10k_ddr_perf_counter_stop(pmu, counter);
+
if (flags & PERF_EF_UPDATE)
cn10k_ddr_perf_event_update(event);
@@ -513,17 +739,19 @@ static void cn10k_ddr_perf_event_del(struct perf_event *event, int flags)
static void cn10k_ddr_perf_pmu_enable(struct pmu *pmu)
{
struct cn10k_ddr_pmu *ddr_pmu = to_cn10k_ddr_pmu(pmu);
+ struct ddr_pmu_platform_data *p_data = ddr_pmu->p_data;
writeq_relaxed(START_OP_CTRL_VAL_START, ddr_pmu->base +
- DDRC_PERF_CNT_START_OP_CTRL);
+ p_data->ddrc_perf_cnt_start_op_ctrl);
}
static void cn10k_ddr_perf_pmu_disable(struct pmu *pmu)
{
struct cn10k_ddr_pmu *ddr_pmu = to_cn10k_ddr_pmu(pmu);
+ struct ddr_pmu_platform_data *p_data = ddr_pmu->p_data;
writeq_relaxed(END_OP_CTRL_VAL_END, ddr_pmu->base +
- DDRC_PERF_CNT_END_OP_CTRL);
+ p_data->ddrc_perf_cnt_end_op_ctrl);
}
static void cn10k_ddr_perf_event_update_all(struct cn10k_ddr_pmu *pmu)
@@ -550,6 +778,7 @@ static void cn10k_ddr_perf_event_update_all(struct cn10k_ddr_pmu *pmu)
static irqreturn_t cn10k_ddr_pmu_overflow_handler(struct cn10k_ddr_pmu *pmu)
{
+ struct ddr_pmu_platform_data *p_data = pmu->p_data;
struct perf_event *event;
struct hw_perf_event *hwc;
u64 prev_count, new_count;
@@ -562,7 +791,8 @@ static irqreturn_t cn10k_ddr_pmu_overflow_handler(struct cn10k_ddr_pmu *pmu)
prev_count = local64_read(&hwc->prev_count);
new_count = cn10k_ddr_perf_read_counter(pmu, hwc->idx);
- /* Overflow condition is when new count less than
+ /*
+ * Overflow condition is when new count less than
* previous count
*/
if (new_count < prev_count)
@@ -575,7 +805,8 @@ static irqreturn_t cn10k_ddr_pmu_overflow_handler(struct cn10k_ddr_pmu *pmu)
prev_count = local64_read(&hwc->prev_count);
new_count = cn10k_ddr_perf_read_counter(pmu, hwc->idx);
- /* Overflow condition is when new count less than
+ /*
+ * Overflow condition is when new count less than
* previous count
*/
if (new_count < prev_count)
@@ -587,11 +818,23 @@ static irqreturn_t cn10k_ddr_pmu_overflow_handler(struct cn10k_ddr_pmu *pmu)
continue;
value = cn10k_ddr_perf_read_counter(pmu, i);
- if (value == DDRC_PERF_CNT_MAX_VALUE) {
+ if (value == p_data->counter_max_val) {
pr_info("Counter-(%d) reached max value\n", i);
- cn10k_ddr_perf_event_update_all(pmu);
- cn10k_ddr_perf_pmu_disable(&pmu->pmu);
- cn10k_ddr_perf_pmu_enable(&pmu->pmu);
+ /*
+ * As separate control register is added for each
+ * counter in odyssey, no need to update all
+ * the events
+ */
+ if (pmu->version == VERSION_V2) {
+ cn10k_ddr_perf_event_update(pmu->events[i]);
+ cn10k_ddr_perf_counter_stop(pmu, i);
+ cn10k_ddr_perf_counter_start(pmu, i);
+
+ } else {
+ cn10k_ddr_perf_event_update_all(pmu);
+ cn10k_ddr_perf_pmu_disable(&pmu->pmu);
+ cn10k_ddr_perf_pmu_enable(&pmu->pmu);
+ }
}
}
@@ -632,7 +875,10 @@ static int cn10k_ddr_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
static int cn10k_ddr_perf_probe(struct platform_device *pdev)
{
+ struct ddr_pmu_platform_data *pltfm_data;
+ struct device *dev = &pdev->dev;
struct cn10k_ddr_pmu *ddr_pmu;
+ const char *compatible;
struct resource *res;
void __iomem *base;
char *name;
@@ -643,6 +889,14 @@ static int cn10k_ddr_perf_probe(struct platform_device *pdev)
return -ENOMEM;
ddr_pmu->dev = &pdev->dev;
+
+ pltfm_data = (struct ddr_pmu_platform_data *)
+ device_get_match_data(&pdev->dev);
+ if (!pltfm_data) {
+ dev_err(&pdev->dev, "Error: No device match data found\n");
+ return -ENODEV;
+ }
+ ddr_pmu->p_data = pltfm_data;
platform_set_drvdata(pdev, ddr_pmu);
base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
@@ -651,25 +905,59 @@ static int cn10k_ddr_perf_probe(struct platform_device *pdev)
ddr_pmu->base = base;
- /* Setup the PMU counter to work in manual mode */
- writeq_relaxed(OP_MODE_CTRL_VAL_MANNUAL, ddr_pmu->base +
- DDRC_PERF_CNT_OP_MODE_CTRL);
-
- ddr_pmu->pmu = (struct pmu) {
- .module = THIS_MODULE,
- .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
- .task_ctx_nr = perf_invalid_context,
- .attr_groups = cn10k_attr_groups,
- .event_init = cn10k_ddr_perf_event_init,
- .add = cn10k_ddr_perf_event_add,
- .del = cn10k_ddr_perf_event_del,
- .start = cn10k_ddr_perf_event_start,
- .stop = cn10k_ddr_perf_event_stop,
- .read = cn10k_ddr_perf_event_update,
- .pmu_enable = cn10k_ddr_perf_pmu_enable,
- .pmu_disable = cn10k_ddr_perf_pmu_disable,
- };
+ ret = device_property_read_string(dev, "compatible", &compatible);
+ if (ret) {
+ pr_err("compatible property not found\n");
+ return ret;
+ }
+
+ if ((strncmp("marvell,cn10k-ddr-pmu", compatible,
+ strlen(compatible)) == 0))
+ ddr_pmu->version = VERSION_V1;
+ else
+ ddr_pmu->version = VERSION_V2;
+
+ if (ddr_pmu->version == VERSION_V1) {
+ ddr_pmu->pmu = (struct pmu) {
+ .module = THIS_MODULE,
+ .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+ .task_ctx_nr = perf_invalid_context,
+ .attr_groups = cn10k_attr_groups,
+ .event_init = cn10k_ddr_perf_event_init,
+ .add = cn10k_ddr_perf_event_add,
+ .del = cn10k_ddr_perf_event_del,
+ .start = cn10k_ddr_perf_event_start,
+ .stop = cn10k_ddr_perf_event_stop,
+ .read = cn10k_ddr_perf_event_update,
+ .pmu_enable = cn10k_ddr_perf_pmu_enable,
+ .pmu_disable = cn10k_ddr_perf_pmu_disable,
+ };
+
+ /*
+ * As we have separate control registers for each counter in Odyssey,
+ * setting up the mode will be done when we enable each counter
+ *
+ */
+ /* Setup the PMU counter to work in manual mode */
+ writeq(OP_MODE_CTRL_VAL_MANUAL, ddr_pmu->base +
+ (ddr_pmu->p_data->ddrc_perf_cnt_op_mode_ctrl));
+ } else {
+ ddr_pmu->pmu = (struct pmu) {
+ .module = THIS_MODULE,
+ .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+ .task_ctx_nr = perf_invalid_context,
+ .attr_groups = odyssey_attr_groups,
+ .event_init = cn10k_ddr_perf_event_init,
+ .add = cn10k_ddr_perf_event_add,
+ .del = cn10k_ddr_perf_event_del,
+ .start = cn10k_ddr_perf_event_start,
+ .stop = cn10k_ddr_perf_event_stop,
+ .read = cn10k_ddr_perf_event_update,
+ .pmu_enable = NULL,
+ .pmu_disable = NULL,
+ };
+ }
/* Choose this cpu to collect perf data */
ddr_pmu->cpu = raw_smp_processor_id();
@@ -712,7 +1000,10 @@ static int cn10k_ddr_perf_remove(struct platform_device *pdev)
#ifdef CONFIG_OF
static const struct of_device_id cn10k_ddr_pmu_of_match[] = {
- { .compatible = "marvell,cn10k-ddr-pmu", },
+ { .compatible = "marvell,cn10k-ddr-pmu",
+ .data = &cn10k_ddr_pmu_pdata },
+ { .compatible = "marvell,odyssey-ddr-pmu",
+ .data = &odyssey_ddr_pmu_pdata },
{ },
};
MODULE_DEVICE_TABLE(of, cn10k_ddr_pmu_of_match);
@@ -720,7 +1011,14 @@ MODULE_DEVICE_TABLE(of, cn10k_ddr_pmu_of_match);
#ifdef CONFIG_ACPI
static const struct acpi_device_id cn10k_ddr_pmu_acpi_match[] = {
- {"MRVL000A", 0},
+ {
+ .id = "MRVL000A",
+ .driver_data = (kernel_ulong_t)&cn10k_ddr_pmu_pdata,
+ },
+ {
+ .id = "MRVL000C",
+ .driver_data = (kernel_ulong_t)&odyssey_ddr_pmu_pdata,
+ },
{},
};
MODULE_DEVICE_TABLE(acpi, cn10k_ddr_pmu_acpi_match);
@@ -752,6 +1050,7 @@ static int __init cn10k_ddr_pmu_init(void)
if (ret)
cpuhp_remove_multi_state(
CPUHP_AP_PERF_ARM_MARVELL_CN10K_DDR_ONLINE);
+
return ret;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH v2 6/6] dt-bindings: perf: Add Marvell Odyssey DDR PMU
2023-08-12 14:25 [PATCH v2 0/6] Marvell Odyssey uncore performance monitor support Gowthami Thiagarajan
` (4 preceding siblings ...)
2023-08-12 14:25 ` [PATCH v2 5/6] perf/marvell: Odyssey DDR Performance monitor support Gowthami Thiagarajan
@ 2023-08-12 14:25 ` Gowthami Thiagarajan
2023-08-14 8:16 ` Krzysztof Kozlowski
5 siblings, 1 reply; 10+ messages in thread
From: Gowthami Thiagarajan @ 2023-08-12 14:25 UTC (permalink / raw)
To: will, mark.rutland, linux-arm-kernel, linux-kernel, robh+dt,
krzysztof.kozlowski+dt, conor+dt, devicetree
Cc: sgoutham, bbhushan2, lcherian, Gowthami Thiagarajan
Add binding documentation for Marvell Odyssey DDR PMU.
Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
---
v1->v2
- No change
.../devicetree/bindings/perf/marvell-cn10k-ddr.yaml | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml b/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
index a18dd0a8c43a..a435cbf4aea0 100644
--- a/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
+++ b/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
@@ -11,10 +11,15 @@ maintainers:
properties:
compatible:
- items:
+ oneOf:
- enum:
- marvell,cn10k-ddr-pmu
-
+ - marvell,odyssey-ddr-pmu
+ - items:
+ - enum:
+ - marvell,cn10k-ddr-pmu
+ - marvell,odyssey-ddr-pmu
+ - const: marvell,cn10k-ddr-pmu
reg:
maxItems: 1
--
2.25.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH v2 6/6] dt-bindings: perf: Add Marvell Odyssey DDR PMU
2023-08-12 14:25 ` [PATCH v2 6/6] dt-bindings: perf: Add Marvell Odyssey DDR PMU Gowthami Thiagarajan
@ 2023-08-14 8:16 ` Krzysztof Kozlowski
0 siblings, 0 replies; 10+ messages in thread
From: Krzysztof Kozlowski @ 2023-08-14 8:16 UTC (permalink / raw)
To: Gowthami Thiagarajan, will, mark.rutland, linux-arm-kernel,
linux-kernel, robh+dt, krzysztof.kozlowski+dt, conor+dt,
devicetree
Cc: sgoutham, bbhushan2, lcherian
On 12/08/2023 16:25, Gowthami Thiagarajan wrote:
> Add binding documentation for Marvell Odyssey DDR PMU.
>
> Signed-off-by: Gowthami Thiagarajan <gthiagarajan@marvell.com>
> ---
>
> v1->v2
> - No change
>
> .../devicetree/bindings/perf/marvell-cn10k-ddr.yaml | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml b/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
> index a18dd0a8c43a..a435cbf4aea0 100644
> --- a/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
> +++ b/Documentation/devicetree/bindings/perf/marvell-cn10k-ddr.yaml
> @@ -11,10 +11,15 @@ maintainers:
>
> properties:
> compatible:
> - items:
> + oneOf:
> - enum:
> - marvell,cn10k-ddr-pmu
> -
> + - marvell,odyssey-ddr-pmu
> + - items:
> + - enum:
> + - marvell,cn10k-ddr-pmu
> + - marvell,odyssey-ddr-pmu
> + - const: marvell,cn10k-ddr-pmu
This is absolutely bogus and does not make sense.
marvell,cn10k-ddr-pmu is not compatible with marvell,cn10k-ddr-pmu
Questions about actual name of SoC remain...
> reg:
> maxItems: 1
>
Best regards,
Krzysztof
^ permalink raw reply [flat|nested] 10+ messages in thread