Re: [PATCH 1/2] drivers: perf: Add LLC-TAD perf counter support

[kajoljain]

On 8/12/21 11:12 PM, Bhaskara Budiredla wrote:
> This driver adds support for Last-level cache tag-and-data unit
> (LLC-TAD) PMU that is featured in some of the Marvell's CN10K
> infrastructure silicons.
> 
> The LLC is divided into 2N slices distributed across N Mesh tiles
> in a single-socket configuration. The driver always configures the
> same counter for all of 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 an application run. The driver does not provide a way
> for the user to partition TADs so that different TADs are used for
> different applications.
> 
> The event counters are zeroed on start event counting to avoid any
> rollover issues. TAD perf counters are 64-bit, so it's not currently
> possible to overflow event counters at current mesh and core
> frequencies.
> 
> To measure tad pmu events use perf tool stat command. For instance:
> 
> perf stat -e tad_dat_msh_in_dss,tad_req_msh_out_any <application>
> perf stat -e tad_alloc_any,tad_hit_any,tad_tag_rd <application>
> 
> Signed-off-by: Bhaskara Budiredla <bbudiredla@xxxxxxxxxxx>
> ---
>  drivers/perf/Kconfig                 |   7 +
>  drivers/perf/Makefile                |   1 +
>  drivers/perf/marvell_cn10k_tad_pmu.c | 429 +++++++++++++++++++++++++++
>  3 files changed, 437 insertions(+)
>  create mode 100644 drivers/perf/marvell_cn10k_tad_pmu.c
> 
> diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
> index 77522e5efe11..73dca11f8080 100644
> --- a/drivers/perf/Kconfig
> +++ b/drivers/perf/Kconfig
> @@ -137,6 +137,13 @@ config ARM_DMC620_PMU
>  	  Support for PMU events monitoring on the ARM DMC-620 memory
>  	  controller.
>  
> +config MARVELL_CN10K_TAD_PMU
> +	tristate "Marvell CN10K LLC-TAD PMU"
> +	depends on ARM64
> +	help
> +	  Provides support for Last-Level cache Tag-and-data Units (LLC-TAD)
> +	  performance monitors on CN10K family silicons.
> +
>  source "drivers/perf/hisilicon/Kconfig"
>  
>  endmenu
> diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
> index 5260b116c7da..2db5418d5b0a 100644
> --- a/drivers/perf/Makefile
> +++ b/drivers/perf/Makefile
> @@ -14,3 +14,4 @@ obj-$(CONFIG_THUNDERX2_PMU) += thunderx2_pmu.o
>  obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o
>  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
> diff --git a/drivers/perf/marvell_cn10k_tad_pmu.c b/drivers/perf/marvell_cn10k_tad_pmu.c
> new file mode 100644
> index 000000000000..fec8b90d52d2
> --- /dev/null
> +++ b/drivers/perf/marvell_cn10k_tad_pmu.c
> @@ -0,0 +1,429 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Marvell CN10K LLC-TAD perf driver
> + *
> + * Copyright (C) 2021 Marvell.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + */
> +
> +#define pr_fmt(fmt) "tad_pmu: " fmt
> +
> +#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>
> +#include <linux/arm-smccc.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 i;
> +
> +	do {
> +		prev = local64_read(&hwc->prev_count);
> +		for (i = 0, new = 0; i < tad_pmu->region_cnt; i++)
> +			new += readq(tad_pmu->regions[i].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 i;
> +
> +	/* TAD()_PFC() stop counting on the write
> +	 * which sets TAD()_PRF()[CNTSEL] == 0
> +	 */
> +	for (i = 0; i < tad_pmu->region_cnt; i++)
> +		writeq_relaxed(0, tad_pmu->regions[i].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 i;
> +
> +	hwc->state = 0;
> +
> +	/* Typically TAD_PFC() are zeroed to start counting */
> +	for (i = 0; i < tad_pmu->region_cnt; i++)
> +		writeq_relaxed(0, tad_pmu->regions[i].base +
> +			       TAD_PFC(counter_idx));
> +
> +	/* TAD()_PFC() start counting on the write
> +	 * which sets TAD()_PRF()[CNTSEL] != 0
> +	 */
> +	for (i = 0; i < tad_pmu->region_cnt; i++) {
> +		reg_val = readq_relaxed(tad_pmu->regions[i].base +
> +					TAD_PRF(counter_idx));
> +		reg_val |= (event_idx & 0xFF);
> +		writeq_relaxed(reg_val,	tad_pmu->regions[i].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.disabled)
> +		return -EINVAL;
> +
> +	if (event->attr.type != event->pmu->type)
> +		return -ENOENT;
> +
> +	if (event->state != PERF_EVENT_STATE_OFF)
> +		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 sprintf(page, "event=0x%02llx\n", pmu_attr->id);
> +}
> +
> +#define TAD_PMU_EVENT_ATTR(name, config)			\
> +	PMU_EVENT_ATTR_ID(name, tad_pmu_event_show, config)
> +
> +static struct attribute *tad_pmu_event_attrs[] = {
> +	TAD_PMU_EVENT_ATTR(tad_none, 0x0),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_in_any, 0x1),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_in_mn, 0x2),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_in_exlmn, 0x3),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_in_any, 0x4),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_in_mn, 0x5),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_in_exlmn, 0x6),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_in_dss, 0x7),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_in_retry_dss, 0x8),
> +	TAD_PMU_EVENT_ATTR(tad_dat_msh_in_any, 0x9),
> +	TAD_PMU_EVENT_ATTR(tad_dat_msh_in_dss, 0xa),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_out_any, 0xb),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_out_dss_rd, 0xc),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_out_dss_wr, 0xd),
> +	TAD_PMU_EVENT_ATTR(tad_req_msh_out_evict, 0xe),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_out_any, 0xf),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_out_retry_exlmn, 0x10),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_out_retry_mn, 0x11),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_out_exlmn, 0x12),
> +	TAD_PMU_EVENT_ATTR(tad_rsp_msh_out_mn, 0x13),
> +	TAD_PMU_EVENT_ATTR(tad_snp_msh_out_any, 0x14),
> +	TAD_PMU_EVENT_ATTR(tad_snp_msh_out_mn, 0x15),
> +	TAD_PMU_EVENT_ATTR(tad_snp_msh_out_exlmn, 0x16),
> +	TAD_PMU_EVENT_ATTR(tad_dat_msh_out_any, 0x17),
> +	TAD_PMU_EVENT_ATTR(tad_dat_msh_out_fill, 0x18),
> +	TAD_PMU_EVENT_ATTR(tad_dat_msh_out_dss, 0x19),
> +	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_dat_rd, 0x21),
> +	TAD_PMU_EVENT_ATTR(tad_dat_rd_byp, 0x22),
> +	TAD_PMU_EVENT_ATTR(tad_ifb_occ, 0x23),
> +	TAD_PMU_EVENT_ATTR(tad_req_occ, 0x24),
> +	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_node *node = pdev->dev.of_node;
> +	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 = of_property_read_u32(node, "tad-page-size", &tad_page_size);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Can't find tad-page-size property\n");
> +		return ret;
> +	}
> +
> +	ret = of_property_read_u32(node, "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 = of_property_read_u32(node, "tad-cnt", &tad_cnt);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Can't find tad-cnt property\n");
> +		return ret;
> +	}
> +
> +	regions = devm_kcalloc(&pdev->dev, 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 |
> +				  PERF_PMU_CAP_NO_INTERRUPT,
> +		.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();
> +

Hi Bhaskara, 
   I think there is some memory leaks in this function. Incase cpuotplug
code fails, don't we need to free momory allocated to `tad_pmu`?

Thanks,
Kajol Jain


> +	/* 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;
> +}
> +
> +static const struct of_device_id tad_pmu_of_match[] = {
> +	{ .compatible = "marvell,cn10k-tad-pmu", },
> +	{},
> +};
> +
> +static struct platform_driver tad_pmu_driver = {
> +	.driver         = {
> +		.name   = "cn10k_tad_pmu",
> +		.of_match_table = of_match_ptr(tad_pmu_of_match),
> +		.suppress_bind_attrs = true,
> +	},
> +	.probe          = tad_pmu_probe,
> +	.remove         = tad_pmu_remove,
> +};
> +
> +static int 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 tad_pmu_init(void)
> +{
> +	int ret;
> +
> +	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> +				      "perf/cn10k/tadpmu:online",
> +				      NULL,
> +				      tad_pmu_offline_cpu);

> +	if (ret < 0)
> +		return ret;
> +	tad_pmu_cpuhp_state = ret;
> +	return platform_driver_register(&tad_pmu_driver);
> +}
> +
> +static void __exit tad_pmu_exit(void)
> +{
> +	platform_driver_unregister(&tad_pmu_driver);
> +	cpuhp_remove_multi_state(tad_pmu_cpuhp_state);
> +}
> +
> +module_init(tad_pmu_init);
> +module_exit(tad_pmu_exit);
> +
> +MODULE_DESCRIPTION("Marvell CN10K LLC-TAD Perf driver");
> +MODULE_AUTHOR("Bhaskara Budiredla <bbudiredla@xxxxxxxxxxx>");
> +MODULE_LICENSE("GPL v2");
>