From patchwork Thu Nov 16 16:23:29 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ji Sheng Teoh X-Patchwork-Id: 13458093 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 9475DC5ACB3 for ; Thu, 16 Nov 2023 16:25:07 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:References:In-Reply-To: Message-ID:Date:Subject:CC:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=FVwAWbdd+MnaYW13YB9qnzHGeFl6HY6zVxd8JJlspis=; b=w9X0Iwi7WTQIV3 pQAYhsMzWBiVvyLUUyGXxctpY8M133dOfgMZI/SDfI6SKCRr0ys62HedCQxkByl7kIxle6mIh2xc6 thXqCBtCxVvQZn2iyVbH4U+CBXTaPPkmEt8vVzj2LvPbRGxDvewsCvIGPtvOJUz28X/XSaVJV3m4l Wtfn/8AHB7u3M7oBxLTeVzLZSBHeXTSpmkazANjESIVua3+SysXrtC53LGf3e8YO2EoVZsZXiz5pw e/BIwkZ9dbX+poEDKteGm+Hy3bMqBDg8nSE+dYIXH4OujM5PhkCC1lHoFrbQ8fcLwb2sBkWsWJrXb EIlKmKuzv3jBFNm03zBQ==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.96 #2 (Red Hat Linux)) id 1r3fAM-003wtH-0F; Thu, 16 Nov 2023 16:24:30 +0000 Received: from ex01.ufhost.com ([61.152.239.75]) by bombadil.infradead.org with esmtp (Exim 4.96 #2 (Red Hat Linux)) id 1r3fAH-003wqy-0C for linux-arm-kernel@lists.infradead.org; Thu, 16 Nov 2023 16:24:28 +0000 Received: from EXMBX166.cuchost.com (unknown [175.102.18.54]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "EXMBX166", Issuer "EXMBX166" (not verified)) by ex01.ufhost.com (Postfix) with ESMTP id 4D6B824DFC2; Fri, 17 Nov 2023 00:24:12 +0800 (CST) Received: from EXMBX172.cuchost.com (172.16.6.92) by EXMBX166.cuchost.com (172.16.6.76) with Microsoft SMTP Server (TLS) id 15.0.1497.42; Fri, 17 Nov 2023 00:24:12 +0800 Received: from localhost.localdomain (202.188.176.82) by EXMBX172.cuchost.com (172.16.6.92) with Microsoft SMTP Server (TLS) id 15.0.1497.42; Fri, 17 Nov 2023 00:24:08 +0800 From: Ji Sheng Teoh To: Will Deacon , Mark Rutland , "Rob Herring" , Krzysztof Kozlowski , Conor Dooley , Thomas Gleixner , Peter Zijlstra CC: Ji Sheng Teoh , Ley Foon Tan , , , Subject: [PATCH v4 1/2] perf: starfive: Add StarLink PMU support Date: Fri, 17 Nov 2023 00:23:29 +0800 Message-ID: <20231116162330.1144983-2-jisheng.teoh@starfivetech.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20231116162330.1144983-1-jisheng.teoh@starfivetech.com> References: <20231116162330.1144983-1-jisheng.teoh@starfivetech.com> MIME-Version: 1.0 X-Originating-IP: [202.188.176.82] X-ClientProxiedBy: EXCAS062.cuchost.com (172.16.6.22) To EXMBX172.cuchost.com (172.16.6.92) X-YovoleRuleAgent: yovoleflag X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20231116_082425_548298_9AB58D1D X-CRM114-Status: GOOD ( 26.08 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org This patch adds support for StarFive's StarLink PMU (Performance Monitor Unit). StarLink PMU integrates one or more CPU cores with a shared L3 memory system. The PMU supports overflow interrupt, up to 16 programmable 64bit event counters, and an independent 64bit cycle counter. StarLink PMU is accessed via MMIO. Example Perf stat output: [root@user]# perf stat -a -e /starfive_starlink_pmu/cycles/ \ -e /starfive_starlink_pmu/read_miss/ \ -e /starfive_starlink_pmu/read_hit/ \ -e /starfive_starlink_pmu/release_request/ \ -e /starfive_starlink_pmu/write_hit/ \ -e /starfive_starlink_pmu/write_miss/ \ -e /starfive_starlink_pmu/write_request/ \ -e /starfive_starlink_pmu/writeback/ \ -e /starfive_starlink_pmu/read_request/ \ -- openssl speed rsa2048 Doing 2048 bits private rsa's for 10s: 5 2048 bits private RSA's in 2.84s Doing 2048 bits public rsa's for 10s: 169 2048 bits public RSA's in 2.42s version: 3.0.11 built on: Tue Sep 19 13:02:31 2023 UTC options: bn(64,64) CPUINFO: N/A sign verify sign/s verify/s rsa 2048 bits 0.568000s 0.014320s 1.8 69.8 ///////// Performance counter stats for 'system wide': 649991998 starfive_starlink_pmu/cycles/ 1009690 starfive_starlink_pmu/read_miss/ 1079750 starfive_starlink_pmu/read_hit/ 2089405 starfive_starlink_pmu/release_request/ 129 starfive_starlink_pmu/write_hit/ 70 starfive_starlink_pmu/write_miss/ 194 starfive_starlink_pmu/write_request/ 150080 starfive_starlink_pmu/writeback/ 2089423 starfive_starlink_pmu/read_request/ 27.062755678 seconds time elapsed Signed-off-by: Ji Sheng Teoh --- drivers/perf/Kconfig | 9 + drivers/perf/Makefile | 1 + drivers/perf/starfive_starlink_pmu.c | 654 +++++++++++++++++++++++++++ include/linux/cpuhotplug.h | 1 + 4 files changed, 665 insertions(+) create mode 100644 drivers/perf/starfive_starlink_pmu.c diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig index 273d67ecf6d2..41278742ef88 100644 --- a/drivers/perf/Kconfig +++ b/drivers/perf/Kconfig @@ -86,6 +86,15 @@ config RISCV_PMU_SBI full perf feature support i.e. counter overflow, privilege mode filtering, counter configuration. +config STARFIVE_STARLINK_PMU + depends on ARCH_STARFIVE + bool "StarFive StarLink PMU" + help + Provide support for StarLink Performance Monitor Unit. + StarLink Performance Monitor Unit integrates one or more cores with + an L3 memory system. The L3 cache events are added into perf event + subsystem, allowing monitoring of various L3 cache perf events. + config ARM_PMU_ACPI depends on ARM_PMU && ACPI def_bool y diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile index 16b3ec4db916..e2153aee1e04 100644 --- a/drivers/perf/Makefile +++ b/drivers/perf/Makefile @@ -15,6 +15,7 @@ obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o obj-$(CONFIG_RISCV_PMU) += riscv_pmu.o obj-$(CONFIG_RISCV_PMU_LEGACY) += riscv_pmu_legacy.o obj-$(CONFIG_RISCV_PMU_SBI) += riscv_pmu_sbi.o +obj-$(CONFIG_STARFIVE_STARLINK_PMU) += starfive_starlink_pmu.o obj-$(CONFIG_THUNDERX2_PMU) += thunderx2_pmu.o obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o obj-$(CONFIG_ARM_SPE_PMU) += arm_spe_pmu.o diff --git a/drivers/perf/starfive_starlink_pmu.c b/drivers/perf/starfive_starlink_pmu.c new file mode 100644 index 000000000000..272896ab1ade --- /dev/null +++ b/drivers/perf/starfive_starlink_pmu.c @@ -0,0 +1,654 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * StarFive's StarLink PMU driver + * + * Copyright (C) 2023 StarFive Technology Co., Ltd. + * + * Author: Ji Sheng Teoh + * + */ + +#define STARLINK_PMU_PDEV_NAME "starfive_starlink_pmu" +#define pr_fmt(fmt) STARLINK_PMU_PDEV_NAME ": " fmt + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define STARLINK_PMU_MAX_COUNTERS 64 +#define STARLINK_PMU_NUM_COUNTERS 16 +#define STARLINK_PMU_IDX_CYCLE_COUNTER 63 + +#define STARLINK_PMU_EVENT_SELECT 0x060 +#define STARLINK_PMU_EVENT_COUNTER 0x160 +#define STARLINK_PMU_COUNTER_MASK GENMASK_ULL(63, 0) +#define STARLINK_PMU_CYCLE_COUNTER 0x058 + +#define STARLINK_PMU_CONTROL 0x040 +#define STARLINK_PMU_GLOBAL_ENABLE BIT(0) + +#define STARLINK_PMU_INTERRUPT_ENABLE 0x050 +#define STARLINK_PMU_COUNTER_OVERFLOW_STATUS 0x048 +#define STARLINK_PMU_CYCLE_OVERFLOW_MASK BIT(63) + +#define CYCLES 0x058 +#define CACHE_READ_REQUEST 0x04000701 +#define CACHE_WRITE_REQUEST 0x03000001 +#define CACHE_RELEASE_REQUEST 0x0003e001 +#define CACHE_READ_HIT 0x00901202 +#define CACHE_READ_MISS 0x04008002 +#define CACHE_WRITE_HIT 0x006c0002 +#define CACHE_WRITE_MISS 0x03000002 +#define CACHE_WRITEBACK 0x00000403 + +#define to_starlink_pmu(p) (container_of(p, struct starlink_pmu, pmu)) + +#define STARLINK_FORMAT_ATTR(_name, _config) \ + (&((struct dev_ext_attribute[]) { \ + { .attr = __ATTR(_name, 0444, starlink_pmu_sysfs_format_show, NULL), \ + .var = (void *)_config, } \ + })[0].attr.attr) + +#define STARLINK_EVENT_ATTR(_name, _id) \ + PMU_EVENT_ATTR_ID(_name, starlink_pmu_sysfs_event_show, _id) + +#define BIT_IS_SET(nr, bit) (((nr) >> (bit)) & 0x1) + +struct starlink_hw_events { + struct perf_event *events[STARLINK_PMU_MAX_COUNTERS]; + DECLARE_BITMAP(used_mask, STARLINK_PMU_MAX_COUNTERS); +}; + +struct starlink_pmu { + struct pmu pmu; + struct starlink_hw_events __percpu *hw_events; + struct hlist_node node; + struct notifier_block starlink_pmu_pm_nb; + void __iomem *pmu_base; + cpumask_t cpumask; + int irq; +}; + +/* Formats Attr */ +static ssize_t +starlink_pmu_sysfs_format_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct dev_ext_attribute *eattr = container_of(attr, + struct dev_ext_attribute, attr); + + return sysfs_emit(buf, "%s\n", (char *)eattr->var); +} + +static struct attribute *starlink_pmu_format_attrs[] = { + STARLINK_FORMAT_ATTR(event, "config:0-31"), + NULL, +}; + +static const struct attribute_group starlink_pmu_format_attr_group = { + .name = "format", + .attrs = starlink_pmu_format_attrs, +}; + +/* Events Attr */ +static ssize_t +starlink_pmu_sysfs_event_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct perf_pmu_events_attr *eattr = container_of(attr, + struct perf_pmu_events_attr, attr); + + return sysfs_emit(buf, "event=0x%02llx\n", eattr->id); +} + +static struct attribute *starlink_pmu_event_attrs[] = { + STARLINK_EVENT_ATTR(cycles, CYCLES), + STARLINK_EVENT_ATTR(read_request, CACHE_READ_REQUEST), + STARLINK_EVENT_ATTR(write_request, CACHE_WRITE_REQUEST), + STARLINK_EVENT_ATTR(release_request, CACHE_RELEASE_REQUEST), + STARLINK_EVENT_ATTR(read_hit, CACHE_READ_HIT), + STARLINK_EVENT_ATTR(read_miss, CACHE_READ_MISS), + STARLINK_EVENT_ATTR(write_hit, CACHE_WRITE_HIT), + STARLINK_EVENT_ATTR(write_miss, CACHE_WRITE_MISS), + STARLINK_EVENT_ATTR(writeback, CACHE_WRITEBACK), + NULL, +}; + +static const struct attribute_group starlink_pmu_events_attr_group = { + .name = "events", + .attrs = starlink_pmu_event_attrs, +}; + +/* Cpumask Attr */ +static ssize_t +cpumask_show(struct device *dev, struct device_attribute *attr, char *buf) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(dev_get_drvdata(dev)); + + return cpumap_print_to_pagebuf(true, buf, &starlink_pmu->cpumask); +} + +static DEVICE_ATTR_RO(cpumask); + +static struct attribute *starlink_pmu_cpumask_attrs[] = { + &dev_attr_cpumask.attr, + NULL, +}; + +static const struct attribute_group starlink_pmu_cpumask_attr_group = { + .attrs = starlink_pmu_cpumask_attrs, +}; + +static const struct attribute_group *starlink_pmu_attr_groups[] = { + &starlink_pmu_format_attr_group, + &starlink_pmu_events_attr_group, + &starlink_pmu_cpumask_attr_group, + NULL, +}; + +static void starlink_pmu_set_event_period(struct perf_event *event) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx = event->hw.idx; + + /* + * Program counter to half of it's max count to handle + * cases of extreme interrupt latency. + */ + u64 val = STARLINK_PMU_COUNTER_MASK >> 1; + + local64_set(&hwc->prev_count, val); + if (hwc->config == CYCLES) + writeq(val, starlink_pmu->pmu_base + STARLINK_PMU_CYCLE_COUNTER); + else + writeq(val, starlink_pmu->pmu_base + STARLINK_PMU_EVENT_COUNTER + + idx * sizeof(u64)); +} + +static void starlink_pmu_counter_start(struct perf_event *event, + struct starlink_pmu *starlink_pmu) +{ + struct hw_perf_event *hwc = &event->hw; + int idx = event->hw.idx; + u64 val; + + /* + * Enable counter overflow interrupt[63:0], + * which is mapped as follow: + * + * event counter 0 - Bit [0] + * event counter 1 - Bit [1] + * ... + * cycle counter - Bit [63] + */ + val = readq(starlink_pmu->pmu_base + STARLINK_PMU_INTERRUPT_ENABLE); + + if (hwc->config == CYCLES) { + /* + * Cycle count has its dedicated register, and it starts + * counting as soon as STARLINK_PMU_GLOBAL_ENABLE is set. + */ + + /* Mask cycle counter overflow interrupt enable bit */ + val |= STARLINK_PMU_CYCLE_OVERFLOW_MASK; + } else { + /* Enable event to monitor through event select register */ + writeq(event->hw.config, starlink_pmu->pmu_base + + STARLINK_PMU_EVENT_SELECT + idx * sizeof(u64)); + + /* Mask event counter overflow interrupt enable bit */ + val |= (1 << idx); + } + + writeq(val, starlink_pmu->pmu_base + STARLINK_PMU_INTERRUPT_ENABLE); + + /* Enable the counter to start counting */ + writeq(STARLINK_PMU_GLOBAL_ENABLE, starlink_pmu->pmu_base + + STARLINK_PMU_CONTROL); +} + +static void starlink_pmu_counter_stop(struct perf_event *event, + struct starlink_pmu *starlink_pmu) +{ + struct hw_perf_event *hwc = &event->hw; + int idx = event->hw.idx; + u64 val; + + /* Stop counter */ + val = readq(starlink_pmu->pmu_base + STARLINK_PMU_CONTROL); + val &= ~STARLINK_PMU_GLOBAL_ENABLE; + writeq(val, starlink_pmu->pmu_base + STARLINK_PMU_CONTROL); + + /* Disable counter overflow interrupt */ + val = readq(starlink_pmu->pmu_base + STARLINK_PMU_INTERRUPT_ENABLE); + if (hwc->config == CYCLES) + val &= ~STARLINK_PMU_CYCLE_OVERFLOW_MASK; + else + val &= ~(1 << idx); + + writeq(val, starlink_pmu->pmu_base + STARLINK_PMU_INTERRUPT_ENABLE); +} + +static void starlink_pmu_update(struct perf_event *event) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + u64 prev_raw_count, new_raw_count; + u64 oldval; + u64 delta; + + do { + prev_raw_count = local64_read(&hwc->prev_count); + if (hwc->config == CYCLES) + new_raw_count = readq(starlink_pmu->pmu_base + + STARLINK_PMU_CYCLE_COUNTER); + else + new_raw_count = readq(starlink_pmu->pmu_base + + STARLINK_PMU_EVENT_COUNTER + + idx * sizeof(u64)); + oldval = local64_cmpxchg(&hwc->prev_count, prev_raw_count, + new_raw_count); + } while (oldval != prev_raw_count); + + delta = (new_raw_count - prev_raw_count) & STARLINK_PMU_COUNTER_MASK; + local64_add(delta, &event->count); +} + +static void starlink_pmu_start(struct perf_event *event, int flags) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + + if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED))) + return; + + if (flags & PERF_EF_RELOAD) + WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE)); + + hwc->state = 0; + + starlink_pmu_set_event_period(event); + starlink_pmu_counter_start(event, starlink_pmu); + + perf_event_update_userpage(event); +} + +static void starlink_pmu_stop(struct perf_event *event, int flags) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + + if (hwc->state & PERF_HES_STOPPED) + return; + + starlink_pmu_counter_stop(event, starlink_pmu); + starlink_pmu_update(event); + hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE; +} + +static int starlink_pmu_add(struct perf_event *event, int flags) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct starlink_hw_events *hw_events = + this_cpu_ptr(starlink_pmu->hw_events); + struct hw_perf_event *hwc = &event->hw; + unsigned long *used_mask = hw_events->used_mask; + u32 n_events = STARLINK_PMU_NUM_COUNTERS; + int idx; + + /* + * Cycle counter has dedicated register to hold counter value. + * Event other than cycle count has to be enabled through + * event select register, and assigned with independent counter + * as they appear. + */ + + if (hwc->config == CYCLES) { + idx = STARLINK_PMU_IDX_CYCLE_COUNTER; + } else { + idx = find_first_zero_bit(used_mask, n_events); + /* All counter are in use */ + if (idx < 0) + return idx; + + set_bit(idx, used_mask); + } + /* Tie counter idx to this event */ + hwc->idx = idx; + hw_events->events[idx] = event; + hwc->state = PERF_HES_UPTODATE | PERF_HES_STOPPED; + + if (flags & PERF_EF_START) + starlink_pmu_start(event, PERF_EF_RELOAD); + + perf_event_update_userpage(event); + + return 0; +} + +static void starlink_pmu_del(struct perf_event *event, int flags) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct starlink_hw_events *hw_events = + this_cpu_ptr(starlink_pmu->hw_events); + struct hw_perf_event *hwc = &event->hw; + + starlink_pmu_stop(event, PERF_EF_UPDATE); + hw_events->events[hwc->idx] = NULL; + clear_bit(hwc->idx, hw_events->used_mask); + + perf_event_update_userpage(event); +} + +static bool starlink_pmu_validate_event_group(struct perf_event *event) +{ + struct perf_event *leader = event->group_leader; + struct perf_event *sibling; + int counter = 1; + + /* + * Ensure hardware events in the group are on the same PMU, + * software events are acceptable. + */ + if (event->group_leader->pmu != event->pmu && + !is_software_event(event->group_leader)) + return false; + + for_each_sibling_event(sibling, leader) { + if (sibling->pmu != event->pmu && !is_software_event(sibling)) + return false; + + counter += 1; + } + /* + * Limit the number of requested counter to + * counter available on the HW. + */ + return counter <= STARLINK_PMU_NUM_COUNTERS; +} + +static int starlink_pmu_event_init(struct perf_event *event) +{ + struct starlink_pmu *starlink_pmu = to_starlink_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + + /* + * Sampling is not supported, as counters are shared + * by all CPU. + */ + if (hwc->sample_period) + return -EOPNOTSUPP; + + /* + * Per-task and attach to a task are not supported, + * as uncore events are not specific to any CPU. + */ + if (event->cpu < 0 || event->attach_state & PERF_ATTACH_TASK) + return -EOPNOTSUPP; + + /* Validate event group */ + if (!starlink_pmu_validate_event_group(event)) + return -EINVAL; + + hwc->idx = -1; + hwc->config = event->attr.config; + event->cpu = cpumask_first(&starlink_pmu->cpumask); + + return 0; +} + +static irqreturn_t starlink_pmu_handle_irq(int irq_num, void *data) +{ + struct starlink_pmu *starlink_pmu = data; + struct starlink_hw_events *hw_events = + this_cpu_ptr(starlink_pmu->hw_events); + bool handled = false; + int idx; + u64 overflow_status; + + for (idx = 0; idx < STARLINK_PMU_MAX_COUNTERS; idx++) { + struct perf_event *event = hw_events->events[idx]; + + overflow_status = readq(starlink_pmu->pmu_base + + STARLINK_PMU_COUNTER_OVERFLOW_STATUS); + if (!BIT_IS_SET(overflow_status, idx)) + continue; + + /* Clear event counter overflow interrupt */ + writeq(1 << idx, starlink_pmu->pmu_base + + STARLINK_PMU_COUNTER_OVERFLOW_STATUS); + + if (!event) + continue; + + starlink_pmu_update(event); + starlink_pmu_set_event_period(event); + handled = true; + } + return IRQ_RETVAL(handled); +} + +static int starlink_setup_irqs(struct starlink_pmu *starlink_pmu, + struct platform_device *pdev) +{ + int ret, irq; + + irq = platform_get_irq(pdev, 0); + if (irq < 0) + return -EINVAL; + + ret = devm_request_irq(&pdev->dev, irq, starlink_pmu_handle_irq, + 0, STARLINK_PMU_PDEV_NAME, starlink_pmu); + if (ret) { + dev_warn(&pdev->dev, "Failed to request IRQ %d\n", irq); + return ret; + } + + starlink_pmu->irq = irq; + + return 0; +} + +#ifdef CONFIG_CPU_PM +static int starlink_pmu_pm_notify(struct notifier_block *b, + unsigned long cmd, void *v) +{ + struct starlink_pmu *starlink_pmu = container_of(b, struct starlink_pmu, + starlink_pmu_pm_nb); + struct starlink_hw_events *hw_events = + this_cpu_ptr(starlink_pmu->hw_events); + int enabled = bitmap_weight(hw_events->used_mask, + STARLINK_PMU_MAX_COUNTERS); + struct perf_event *event; + int idx; + + if (!enabled) + return NOTIFY_OK; + + for (idx = 0; idx < STARLINK_PMU_MAX_COUNTERS; idx++) { + event = hw_events->events[idx]; + if (!event) + continue; + + switch (cmd) { + case CPU_PM_ENTER: + /* Stop and update the counter */ + starlink_pmu_stop(event, PERF_EF_UPDATE); + break; + case CPU_PM_EXIT: + case CPU_PM_ENTER_FAILED: + /* Restore and enable the counter */ + starlink_pmu_start(event, PERF_EF_RELOAD); + break; + default: + break; + } + } + + return NOTIFY_OK; +} + +static int starlink_pmu_pm_register(struct starlink_pmu *starlink_pmu) +{ + starlink_pmu->starlink_pmu_pm_nb.notifier_call = starlink_pmu_pm_notify; + return cpu_pm_register_notifier(&starlink_pmu->starlink_pmu_pm_nb); +} + +static void starlink_pmu_pm_unregister(struct starlink_pmu *starlink_pmu) +{ + cpu_pm_unregister_notifier(&starlink_pmu->starlink_pmu_pm_nb); +} +#else +static inline int +starlink_pmu_pm_register(struct starlink_pmu *starlink_pmu) { return 0; } +static inline void +starlink_pmu_pm_unregister(struct starlink_pmu *starlink_pmu) { } +#endif + +static void starlink_pmu_destroy(struct starlink_pmu *starlink_pmu) +{ + starlink_pmu_pm_unregister(starlink_pmu); + cpuhp_state_remove_instance(CPUHP_AP_PERF_RISCV_STARFIVE_STARLINK_ONLINE, + &starlink_pmu->node); +} + +static int starlink_pmu_probe(struct platform_device *pdev) +{ + struct starlink_pmu *starlink_pmu; + struct starlink_hw_events *hw_events; + struct resource *res; + int cpuid, i, ret; + + starlink_pmu = devm_kzalloc(&pdev->dev, sizeof(*starlink_pmu), GFP_KERNEL); + if (!starlink_pmu) + return -ENOMEM; + + starlink_pmu->pmu_base = + devm_platform_get_and_ioremap_resource(pdev, 0, &res); + if (IS_ERR(starlink_pmu->pmu_base)) + return PTR_ERR(starlink_pmu->pmu_base); + + ret = starlink_setup_irqs(starlink_pmu, pdev); + + ret = cpuhp_state_add_instance(CPUHP_AP_PERF_RISCV_STARFIVE_STARLINK_ONLINE, + &starlink_pmu->node); + if (ret) + return ret; + + ret = starlink_pmu_pm_register(starlink_pmu); + if (ret) + starlink_pmu_destroy(starlink_pmu); + + starlink_pmu->hw_events = alloc_percpu_gfp(struct starlink_hw_events, + GFP_KERNEL); + if (!starlink_pmu->hw_events) { + pr_info("Failed to allocate per-cpu PMU data.\n"); + kfree(starlink_pmu); + return -ENOMEM; + } + + for_each_possible_cpu(cpuid) { + hw_events = per_cpu_ptr(starlink_pmu->hw_events, cpuid); + for (i = 0; i < STARLINK_PMU_MAX_COUNTERS; i++) + hw_events->events[i] = NULL; + } + + starlink_pmu->pmu = (struct pmu) { + .task_ctx_nr = perf_invalid_context, + .event_init = starlink_pmu_event_init, + .add = starlink_pmu_add, + .del = starlink_pmu_del, + .start = starlink_pmu_start, + .stop = starlink_pmu_stop, + .read = starlink_pmu_update, + .attr_groups = starlink_pmu_attr_groups, + }; + + ret = perf_pmu_register(&starlink_pmu->pmu, STARLINK_PMU_PDEV_NAME, -1); + if (ret) + starlink_pmu_destroy(starlink_pmu); + + dev_info(&pdev->dev, "Registered StarFive's StarLink PMU\n"); + + return ret; +} + +static const struct of_device_id starlink_pmu_of_match[] = { + { .compatible = "starfive,jh8100-starlink-pmu", }, + {}, +}; +MODULE_DEVICE_TABLE(of, starlink_pmu_of_match); + +static struct platform_driver starlink_pmu_driver = { + .driver = { + .name = STARLINK_PMU_PDEV_NAME, + .of_match_table = starlink_pmu_of_match, + .suppress_bind_attrs = true, + }, + .probe = starlink_pmu_probe, +}; + +static int +starlink_pmu_online_cpu(unsigned int cpu, struct hlist_node *node) +{ + struct starlink_pmu *starlink_pmu = hlist_entry_safe(node, + struct starlink_pmu, + node); + + if (cpumask_empty(&starlink_pmu->cpumask)) + cpumask_set_cpu(cpu, &starlink_pmu->cpumask); + + WARN_ON(irq_set_affinity(starlink_pmu->irq, cpumask_of(cpu))); + + return 0; +} + +static int +starlink_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node) +{ + struct starlink_pmu *starlink_pmu = hlist_entry_safe(node, + struct starlink_pmu, + node); + unsigned int target; + + if (!cpumask_test_and_clear_cpu(cpu, &starlink_pmu->cpumask)) + return 0; + + target = cpumask_any_but(cpu_online_mask, cpu); + if (target >= nr_cpu_ids) + return 0; + + perf_pmu_migrate_context(&starlink_pmu->pmu, cpu, target); + + cpumask_set_cpu(target, &starlink_pmu->cpumask); + WARN_ON(irq_set_affinity(starlink_pmu->irq, cpumask_of(target))); + + return 0; +} + +static int __init starlink_pmu_init(void) +{ + int ret; + + ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_RISCV_STARFIVE_STARLINK_ONLINE, + "soc/starfive/starlink_pmu:online", + starlink_pmu_online_cpu, + starlink_pmu_offline_cpu); + if (ret) + return ret; + + return platform_driver_register(&starlink_pmu_driver); +} + +device_initcall(starlink_pmu_init); diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index d305db70674b..6d9eb70c13d4 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -219,6 +219,7 @@ enum cpuhp_state { CPUHP_AP_PERF_X86_CQM_ONLINE, CPUHP_AP_PERF_X86_CSTATE_ONLINE, CPUHP_AP_PERF_X86_IDXD_ONLINE, + CPUHP_AP_PERF_RISCV_STARFIVE_STARLINK_ONLINE, CPUHP_AP_PERF_S390_CF_ONLINE, CPUHP_AP_PERF_S390_SF_ONLINE, CPUHP_AP_PERF_ARM_CCI_ONLINE, From patchwork Thu Nov 16 16:23:30 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ji Sheng Teoh X-Patchwork-Id: 13458095 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id DCAD7C197A0 for ; Thu, 16 Nov 2023 16:25:50 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:References:In-Reply-To: Message-ID:Date:Subject:CC:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=E/o7vRuE/bHlsTp7zverl+oYcDzqQeUKma+XXJG6DOw=; b=UWt2bSGygs6TLd aMeuSzX9NvfjQJqgzGZFXt+naTQH6+E/PNRuSFW0pYSrPTz00dbucq8IySD8GNp5kUUjuqq5jhd8v Q6tLLkOwqzk5ud/A752K+Y3R+2YB/UGMhukzyjPz8LfJSF6jrynlAkgCEtv9Qnlm3eiTd2sNtyQTs 0sQHsKP4OsC3qScWd8D2D8OlVASpxO/mrzx7z4987mMpAc5VIcSCJhmnWZIcwYFp9aRZJHR9VjPMI IxTyEq83YKnmRvp23TniqW2BWGddR0oRE0HiJPrkqAqjhwo3zeOXWqrtDaK9I9AVYOejhZ4Oc4I5P K50HV4EHJNz8qaUe20HQ==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.96 #2 (Red Hat Linux)) id 1r3fBF-003xD2-11; Thu, 16 Nov 2023 16:25:25 +0000 Received: from desiato.infradead.org ([2001:8b0:10b:1:d65d:64ff:fe57:4e05]) by bombadil.infradead.org with esmtps (Exim 4.96 #2 (Red Hat Linux)) id 1r3fBD-003xCB-2C for linux-arm-kernel@bombadil.infradead.org; Thu, 16 Nov 2023 16:25:23 +0000 DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=desiato.20200630; h=Content-Transfer-Encoding:Content-Type :MIME-Version:References:In-Reply-To:Message-ID:Date:Subject:CC:To:From: Sender:Reply-To:Content-ID:Content-Description; bh=wkB56k+YEyGYeAITM0zC3CCEgDGHquiJ2Vzw18gIXVQ=; b=nEVNikbDo8XSOa8beRHztwZzDL W1yX7OTABMBDsISTWsiE5dZvu1GporE4jItET9CLppC3jEmRVVDZyHHm4WXGK+QTUt49SzeriLLcy vLVbt8xCQoUYTUuGUaeA0RVjB8RTCZy+Ah3dtXaXeME7ayl0SzwGU4AFcfRwLkz6ZVmJZcPj4P/PG kHy1V5WbKZHZCqSW/RPfB8Z8qll3+/EEplB6SgFmop818iK9GKr2moVWXRUe3fxQVIM2L7zh1FQcH bzRbmFq2Bh8HHOIGSp6CUPOyiHvC/AoDvbzjmMESOnE6ksuV+UNK62za2evXNPdYHtqFpgVErXGPQ iFqx79CA==; Received: from ex01.ufhost.com ([61.152.239.75]) by desiato.infradead.org with esmtp (Exim 4.96 #2 (Red Hat Linux)) id 1r3fB9-005g9f-1J for linux-arm-kernel@lists.infradead.org; Thu, 16 Nov 2023 16:25:22 +0000 Received: from EXMBX165.cuchost.com (unknown [175.102.18.54]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "EXMBX165", Issuer "EXMBX165" (not verified)) by ex01.ufhost.com (Postfix) with ESMTP id 3122024E083; Fri, 17 Nov 2023 00:24:16 +0800 (CST) Received: from EXMBX172.cuchost.com (172.16.6.92) by EXMBX165.cuchost.com (172.16.6.75) with Microsoft SMTP Server (TLS) id 15.0.1497.42; Fri, 17 Nov 2023 00:24:16 +0800 Received: from localhost.localdomain (202.188.176.82) by EXMBX172.cuchost.com (172.16.6.92) with Microsoft SMTP Server (TLS) id 15.0.1497.42; Fri, 17 Nov 2023 00:24:11 +0800 From: Ji Sheng Teoh To: Will Deacon , Mark Rutland , "Rob Herring" , Krzysztof Kozlowski , Conor Dooley , Thomas Gleixner , Peter Zijlstra CC: Ji Sheng Teoh , Ley Foon Tan , , , Subject: [PATCH v4 2/2] dt-bindings: perf: starfive: Add JH8100 StarLink PMU Date: Fri, 17 Nov 2023 00:23:30 +0800 Message-ID: <20231116162330.1144983-3-jisheng.teoh@starfivetech.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20231116162330.1144983-1-jisheng.teoh@starfivetech.com> References: <20231116162330.1144983-1-jisheng.teoh@starfivetech.com> MIME-Version: 1.0 X-Originating-IP: [202.188.176.82] X-ClientProxiedBy: EXCAS062.cuchost.com (172.16.6.22) To EXMBX172.cuchost.com (172.16.6.92) X-YovoleRuleAgent: yovoleflag X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20231116_162520_154898_78A088B9 X-CRM114-Status: GOOD ( 12.36 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org Add device tree binding for StarFive's JH8100 StarLink PMU (Performance Monitor Unit). Signed-off-by: Ji Sheng Teoh Reviewed-by: Conor Dooley --- .../perf/starfive,jh8100-starlink-pmu.yaml | 46 +++++++++++++++++++ 1 file changed, 46 insertions(+) create mode 100644 Documentation/devicetree/bindings/perf/starfive,jh8100-starlink-pmu.yaml diff --git a/Documentation/devicetree/bindings/perf/starfive,jh8100-starlink-pmu.yaml b/Documentation/devicetree/bindings/perf/starfive,jh8100-starlink-pmu.yaml new file mode 100644 index 000000000000..915c6b814026 --- /dev/null +++ b/Documentation/devicetree/bindings/perf/starfive,jh8100-starlink-pmu.yaml @@ -0,0 +1,46 @@ +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/perf/starfive,jh8100-starlink-pmu.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: StarFive JH8100 StarLink PMU + +maintainers: + - Ji Sheng Teoh + +description: + StarFive's JH8100 StarLink PMU integrates one or more CPU cores with a + shared L3 memory system. The PMU support overflow interrupt, up to + 16 programmable 64bit event counters, and an independent 64bit cycle + counter. StarFive's JH8100 StarLink PMU is accessed via MMIO. + +properties: + compatible: + const: starfive,jh8100-starlink-pmu + + reg: + maxItems: 1 + + interrupts: + maxItems: 1 + +required: + - compatible + - reg + - interrupts + +additionalProperties: false + +examples: + - | + soc { + #address-cells = <2>; + #size-cells = <2>; + + pmu@12900000 { + compatible = "starfive,jh8100-starlink-pmu"; + reg = <0x0 0x12900000 0x0 0x10000>; + interrupts = <34>; + }; + };