All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH V12 1/4] dt-bindings: perf: imx8-ddr: add imx8qxp ddr performance monitor
@ 2019-05-01 18:43 ` Frank Li
  0 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Added binding doc for imx8qxp ddr performance monitor

Signed-off-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Rob Herring <robh@kernel.org>
---

Notes:
    No change from v10 to v12
    
    Change from v8 to v9
    * use 32bit address width
    
    No change from v4 to v8
    
    Change from v4 to v4
    * remove "standard xxx"
    
    Change from v2 to v3
    * ddr_pmu0 -> ddr-pmu

 .../devicetree/bindings/perf/fsl-imx-ddr.txt       | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt

diff --git a/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt b/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
new file mode 100644
index 0000000..9b9cda6a
--- /dev/null
+++ b/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
@@ -0,0 +1,22 @@
+* Freescale(NXP) IMX8 DDR performance monitor
+
+Required properties:
+
+- compatible: should be one of:
+	"fsl,imx8-ddr-pmu"
+	"fsl,imx8m-ddr-pmu"
+
+- reg: physical address and size
+
+- interrupts: single interrupt
+	generated by the control block
+
+Example:
+
+	ddr-pmu@5c020000 {
+		compatible = "fsl,imx8-ddr-pmu";
+		reg = <0x5c020000 0x10000>;
+		interrupt-parent = <&gic>;
+		interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+	};
+
-- 
2.5.2

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 1/4] dt-bindings: perf: imx8-ddr: add imx8qxp ddr performance monitor
@ 2019-05-01 18:43 ` Frank Li
  0 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Added binding doc for imx8qxp ddr performance monitor

Signed-off-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Rob Herring <robh@kernel.org>
---

Notes:
    No change from v10 to v12
    
    Change from v8 to v9
    * use 32bit address width
    
    No change from v4 to v8
    
    Change from v4 to v4
    * remove "standard xxx"
    
    Change from v2 to v3
    * ddr_pmu0 -> ddr-pmu

 .../devicetree/bindings/perf/fsl-imx-ddr.txt       | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt

diff --git a/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt b/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
new file mode 100644
index 0000000..9b9cda6a
--- /dev/null
+++ b/Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
@@ -0,0 +1,22 @@
+* Freescale(NXP) IMX8 DDR performance monitor
+
+Required properties:
+
+- compatible: should be one of:
+	"fsl,imx8-ddr-pmu"
+	"fsl,imx8m-ddr-pmu"
+
+- reg: physical address and size
+
+- interrupts: single interrupt
+	generated by the control block
+
+Example:
+
+	ddr-pmu@5c020000 {
+		compatible = "fsl,imx8-ddr-pmu";
+		reg = <0x5c020000 0x10000>;
+		interrupt-parent = <&gic>;
+		interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+	};
+
-- 
2.5.2

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-05-01 18:43 ` Frank Li
@ 2019-05-01 18:43   ` Frank Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add ddr performance monitor support for iMX8QXP

There are 4 counters for ddr perfomance events.
counter 0 is dedicated for cycles.
you choose any up to 3 no cycles events.

for example:

perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls

Support below events.

  imx8_ddr0/activate/                                     [Kernel PMU event]
  imx8_ddr0/axid-read/                                    [Kernel PMU event]
  imx8_ddr0/axid-write/                                   [Kernel PMU event]
  imx8_ddr0/cycles/                                       [Kernel PMU event]
  imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
  imx8_ddr0/hp-read/                                      [Kernel PMU event]
  imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
  imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/load-mode/                                    [Kernel PMU event]
  imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
  imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
  imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/mwr/                                          [Kernel PMU event]
  imx8_ddr0/precharge/                                    [Kernel PMU event]
  imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
  imx8_ddr0/read-access/                                  [Kernel PMU event]
  imx8_ddr0/read-activate/                                [Kernel PMU event]
  imx8_ddr0/read-command/                                 [Kernel PMU event]
  imx8_ddr0/read-cycles/                                  [Kernel PMU event]
  imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
  imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
  imx8_ddr0/read-write-transition/                        [Kernel PMU event]
  imx8_ddr0/read/                                         [Kernel PMU event]
  imx8_ddr0/refresh/                                      [Kernel PMU event]
  imx8_ddr0/selfresh/                                     [Kernel PMU event]
  imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/write-access/                                 [Kernel PMU event]
  imx8_ddr0/write-command/                                [Kernel PMU event]
  imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
  imx8_ddr0/write-cycles/                                 [Kernel PMU event]
  imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
  imx8_ddr0/write/

Signed-off-by: Frank Li <Frank.Li@nxp.com>
---

Notes:
    Change from v11 to v12
    * change name from imx_ddr to imx8_ddr

    Change from v10 to v11
     * using devm_platform_ioremap_resource
     * using help macro IMX8_DDR_PMU_EVENT_ATTR

    Change from v9 to v10
     * use name imx8_ddr_perf_pmu
     * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
     * config: 0-7
     * use read_relaxed() at ddr_perf_read_counter
     * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
     * raname ddr_perf_event_start to ddr_perf_counter_enable
     * use FIELD_PREP
     * write 0 to disalbe counter at ddr_perf_event_enable
     * update hwc->state
     * remove local64_set at ddr_perf_event_add
     * move perf_pmu_register after irq_register to simplfy err handle

    No change from v8 to v9

    Change from v7 to v8
     * remove unused define
     * change total_events to active_events, change active_events to events
     * remove flags,
     * fix multi line comments code sytle
     * add pmu_enable\disable function
     * disable event at irq handle
     * remove counter check at ddr_perf_free_counter
     * remove pmu->irq check
     * add group check

    Change from v6 to v7
     * added irq affinity handle, ref arm-ccn.c
     * added IRQF_NOBALANCING | IRQF_NO_THREAD
     * added ida_simple_remove at failure path

    Change from v5 to v6
     * fix insmod\rmmod problem
     * remove randunt register read at irq handle
     * change u32 irq to int
     * devm_request_irq use default flags.

    Change from v4 to v5
     * Remove AXI ID filter function

    Change from v3 to v4
     * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
     * sort include
     * remove struct fsl_ddr_devtype_data
     * Added comment need disable control first
     * Added comment about must enable cycle counter
     * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
     * Added comment about cycle counter is fastest one

    Change from v2 to v3
     * remove kfree

    Change from V1 to V2
     * update Kconfig by use i.MX8 instead of i.MX8 QXP
     * remove gpl statememnt since SPDX tag
     * use dev_kzalloc
     * use dev_err
     * commit message show axi_read 0x41\axi_write 0x42
     * commit message show cycles must be enabled
     * Irq only issue at cycles overflow
     * use NUM_COUNTER
     * use devm_request_irq
     * add hotplug callback to handle context migration

 drivers/perf/Kconfig             |   7 +
 drivers/perf/Makefile            |   1 +
 drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 560 insertions(+)
 create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c

diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index a94e586..9bc3785 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -70,6 +70,13 @@ config ARM_DSU_PMU
 	  system, control logic. The PMU allows counting various events related
 	  to DSU.
 
+config FSL_IMX8_DDR_PMU
+	tristate "Freescale i.MX8 DDR perf monitor"
+	depends on ARCH_MXC
+	  help
+	  Provides support for ddr perfomance monitor in i.MX8. Provide memory
+	  througput information.
+
 config HISI_PMU
        bool "HiSilicon SoC PMU"
        depends on ARM64 && ACPI
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index 3048994..2ebb4de 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
 obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
 obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
 obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
+obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
 obj-$(CONFIG_HISI_PMU) += hisilicon/
 obj-$(CONFIG_QCOM_L2_PMU)	+= qcom_l2_pmu.o
 obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
new file mode 100644
index 0000000..15b0772
--- /dev/null
+++ b/drivers/perf/fsl_imx8_ddr_perf.c
@@ -0,0 +1,552 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2017 NXP
+ * Copyright 2016 Freescale Semiconductor, Inc.
+ */
+
+#include <linux/bitfield.h>
+#include <linux/init.h>
+#include <linux/interrupt.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/of_irq.h>
+#include <linux/perf_event.h>
+#include <linux/slab.h>
+
+#define COUNTER_CNTL		0x0
+#define COUNTER_READ		0x20
+
+#define COUNTER_DPCR1		0x30
+
+#define CNTL_OVER		0x1
+#define CNTL_CLEAR		0x2
+#define CNTL_EN			0x4
+#define CNTL_EN_MASK		0xFFFFFFFB
+#define CNTL_CLEAR_MASK		0xFFFFFFFD
+#define CNTL_OVER_MASK		0xFFFFFFFE
+
+#define CNTL_CSV_SHIFT		24
+#define CNTL_CSV_MASK		(0xFF << CNTL_CSV_SHIFT)
+
+#define EVENT_CYCLES_ID		0
+#define EVENT_CYCLES_COUNTER	0
+#define NUM_COUNTERS		4
+
+#define to_ddr_pmu(p)		container_of(p, struct ddr_pmu, pmu)
+
+#define DDR_PERF_DEV_NAME	"imx8_ddr_perf_pmu"
+
+static DEFINE_IDA(ddr_ida);
+
+static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
+	{ .compatible = "fsl,imx8-ddr-pmu",},
+	{ .compatible = "fsl,imx8m-ddr-pmu",},
+	{ /* sentinel */ }
+};
+
+struct ddr_pmu {
+	struct pmu pmu;
+	void __iomem *base;
+	unsigned int cpu;
+	struct	hlist_node node;
+	struct	device *dev;
+	struct perf_event *events[NUM_COUNTERS];
+	int active_events;
+	enum cpuhp_state cpuhp_state;
+	int irq;
+	int id;
+};
+
+static ssize_t ddr_perf_cpumask_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct ddr_pmu *pmu = dev_get_drvdata(dev);
+
+	return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
+}
+
+static struct device_attribute ddr_perf_cpumask_attr =
+	__ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
+
+static struct attribute *ddr_perf_cpumask_attrs[] = {
+	&ddr_perf_cpumask_attr.attr,
+	NULL,
+};
+
+static struct attribute_group ddr_perf_cpumask_attr_group = {
+	.attrs = ddr_perf_cpumask_attrs,
+};
+
+static ssize_t
+ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)				\
+	(&((struct perf_pmu_events_attr[]) {				\
+		{ .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
+		  .id = _id, }						\
+	})[0].attr.attr)
+
+static struct attribute *ddr_perf_events_attrs[] = {
+	IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
+	IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
+	IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
+	IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
+	IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
+	IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
+	IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
+	IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
+	IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
+	IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
+	IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
+	IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
+	IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
+	IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
+	IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
+	IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
+	IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
+	IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
+	IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
+	IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
+	IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
+	IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
+	IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
+	NULL,
+};
+
+static struct attribute_group ddr_perf_events_attr_group = {
+	.name = "events",
+	.attrs = ddr_perf_events_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-7");
+
+static struct attribute *ddr_perf_format_attrs[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group ddr_perf_format_attr_group = {
+	.name = "format",
+	.attrs = ddr_perf_format_attrs,
+};
+
+static const struct attribute_group *attr_groups[] = {
+	&ddr_perf_events_attr_group,
+	&ddr_perf_format_attr_group,
+	&ddr_perf_cpumask_attr_group,
+	NULL,
+};
+
+static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
+{
+	int i;
+
+	/*
+	 * Always map cycle event to counter 0
+	 * Cycles counter is dedicated for cycle event
+	 * can't used for the other events
+	 */
+	if (event == EVENT_CYCLES_ID) {
+		if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+			return EVENT_CYCLES_COUNTER;
+		else
+			return -ENOENT;
+	}
+
+	for (i = 1; i < NUM_COUNTERS; i++) {
+		if (pmu->events[i] == NULL)
+			return i;
+	}
+
+	return -ENOENT;
+}
+
+static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
+{
+	pmu->events[counter] = NULL;
+}
+
+static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
+{
+	return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
+}
+
+static int ddr_perf_event_init(struct perf_event *event)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	struct perf_event *sibling;
+
+	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) {
+		dev_warn(pmu->dev, "Can't provide per-task data!\n");
+		return -EOPNOTSUPP;
+	}
+
+	/*
+	 * We must NOT create groups containing mixed PMUs, although software
+	 * events are acceptable (for example to create a CCN group
+	 * periodically read when a hrtimer aka cpu-clock leader triggers).
+	 */
+	if (event->group_leader->pmu != event->pmu &&
+			!is_software_event(event->group_leader))
+		return -EINVAL;
+
+	for_each_sibling_event(sibling, event->group_leader) {
+		if (sibling->pmu != event->pmu &&
+				!is_software_event(sibling))
+			return -EINVAL;
+	}
+
+	event->cpu = pmu->cpu;
+	hwc->idx = -1;
+
+	return 0;
+}
+
+
+static void ddr_perf_event_update(struct perf_event *event)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	u64 delta, prev_raw_count, new_raw_count;
+	int counter = hwc->idx;
+
+	do {
+		prev_raw_count = local64_read(&hwc->prev_count);
+		new_raw_count = ddr_perf_read_counter(pmu, counter);
+	} while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+			new_raw_count) != prev_raw_count);
+
+	delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
+
+	local64_add(delta, &event->count);
+}
+
+static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
+				  int counter, bool enable)
+{
+	u8 reg = counter * 4 + COUNTER_CNTL;
+	int val;
+
+	if (enable) {
+		/*
+		 * must disable first, then enable again
+		 * otherwise, cycle counter will not work
+		 * if previous state is enabled.
+		 */
+		writel(0, pmu->base + reg);
+		val = CNTL_EN | CNTL_CLEAR;
+		val |= FIELD_PREP(CNTL_CSV_MASK, config);
+		writel(val, pmu->base + reg);
+	} else {
+		/* Disable counter */
+		writel(0, pmu->base + reg);
+	}
+}
+
+static void ddr_perf_event_start(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	local64_set(&hwc->prev_count, 0);
+
+	ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
+
+	hwc->state = 0;
+}
+
+static int ddr_perf_event_add(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter;
+	int cfg = event->attr.config;
+
+	counter = ddr_perf_alloc_counter(pmu, cfg);
+	if (counter < 0) {
+		dev_dbg(pmu->dev, "There are not enough counters\n");
+		return -EOPNOTSUPP;
+	}
+
+	pmu->events[counter] = event;
+	pmu->active_events++;
+	hwc->idx = counter;
+
+	hwc->state |= PERF_HES_STOPPED;
+
+	if (flags & PERF_EF_START)
+		ddr_perf_event_start(event, flags);
+
+	return 0;
+}
+
+static void ddr_perf_event_stop(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
+	ddr_perf_event_update(event);
+
+	hwc->state |= PERF_HES_STOPPED;
+}
+
+static void ddr_perf_event_del(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	ddr_perf_event_stop(event, PERF_EF_UPDATE);
+
+	ddr_perf_free_counter(pmu, counter);
+	pmu->active_events--;
+	hwc->idx = -1;
+}
+
+static void ddr_perf_pmu_enable(struct pmu *pmu)
+{
+	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
+
+	/* enable cycle counter if cycle is not active event list */
+	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+		ddr_perf_counter_enable(ddr_pmu,
+				      EVENT_CYCLES_ID,
+				      EVENT_CYCLES_COUNTER,
+				      true);
+}
+
+static void ddr_perf_pmu_disable(struct pmu *pmu)
+{
+	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
+
+	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+		ddr_perf_counter_enable(ddr_pmu,
+				      EVENT_CYCLES_ID,
+				      EVENT_CYCLES_COUNTER,
+				      false);
+}
+
+static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
+			 struct device *dev)
+{
+	*pmu = (struct ddr_pmu) {
+		.pmu = (struct pmu) {
+			.capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+			.task_ctx_nr = perf_invalid_context,
+			.attr_groups = attr_groups,
+			.event_init  = ddr_perf_event_init,
+			.add	     = ddr_perf_event_add,
+			.del	     = ddr_perf_event_del,
+			.start	     = ddr_perf_event_start,
+			.stop	     = ddr_perf_event_stop,
+			.read	     = ddr_perf_event_update,
+			.pmu_enable  = ddr_perf_pmu_enable,
+			.pmu_disable = ddr_perf_pmu_disable,
+		},
+		.base = base,
+		.dev = dev,
+	};
+
+	pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
+	return pmu->id;
+}
+
+static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
+{
+	int i;
+	struct ddr_pmu *pmu = (struct ddr_pmu *) p;
+	struct perf_event *event, *cycle_event = NULL;
+
+	/* all counter will stop if cycle counter disabled */
+	ddr_perf_counter_enable(pmu,
+			      EVENT_CYCLES_ID,
+			      EVENT_CYCLES_COUNTER,
+			      false);
+	/*
+	 * When the cycle counter overflows, all counters are stopped,
+	 * and an IRQ is raised. If any other counter overflows, it
+	 * continues counting, and no IRQ is raised.
+	 *
+	 * Cycles occur at least 4 times as often as other events, so we
+	 * can update all events on a cycle counter overflow and not
+	 * lose events.
+	 *
+	 */
+	for (i = 0; i < NUM_COUNTERS; i++) {
+
+		if (!pmu->events[i])
+			continue;
+
+		event = pmu->events[i];
+
+		ddr_perf_event_update(event);
+
+		if (event->hw.idx == EVENT_CYCLES_COUNTER)
+			cycle_event = event;
+	}
+
+	ddr_perf_counter_enable(pmu,
+			      EVENT_CYCLES_ID,
+			      EVENT_CYCLES_COUNTER,
+			      true);
+	if (cycle_event)
+		ddr_perf_event_update(cycle_event);
+
+	return IRQ_HANDLED;
+}
+
+static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+	struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
+	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;
+
+	WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
+
+	return 0;
+}
+
+static int ddr_perf_probe(struct platform_device *pdev)
+{
+	struct ddr_pmu *pmu;
+	struct device_node *np;
+	void __iomem *base;
+	char *name;
+	int num;
+	int ret;
+	int irq;
+
+	base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	np = pdev->dev.of_node;
+
+	pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
+	if (!pmu)
+		return -ENOMEM;
+
+	num = ddr_perf_init(pmu, base, &pdev->dev);
+
+	platform_set_drvdata(pdev, pmu);
+
+	name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
+	if (!name)
+		return -ENOMEM;
+
+	pmu->cpu = raw_smp_processor_id();
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
+				      DDR_PERF_DEV_NAME,
+				      NULL,
+				      ddr_perf_offline_cpu);
+
+	if (ret < 0) {
+		dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
+		goto ddr_perf_err;
+	}
+
+	pmu->cpuhp_state = ret;
+
+	/* Register the pmu instance for cpu hotplug */
+	cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+
+	/* Request irq */
+	irq = of_irq_get(np, 0);
+	if (irq < 0) {
+		dev_err(&pdev->dev, "Failed to get irq: %d", irq);
+		ret = irq;
+		goto ddr_perf_err;
+	}
+
+	ret = devm_request_irq(&pdev->dev, irq,
+					ddr_perf_irq_handler,
+					IRQF_NOBALANCING | IRQF_NO_THREAD,
+					DDR_PERF_DEV_NAME,
+					pmu);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "Request irq failed: %d", ret);
+		goto ddr_perf_err;
+	}
+
+	pmu->irq = irq;
+	ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
+	if (ret) {
+		dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
+		goto ddr_perf_err;
+	}
+
+	ret = perf_pmu_register(&pmu->pmu, name, -1);
+	if (ret)
+		goto ddr_perf_err;
+
+	return 0;
+
+ddr_perf_err:
+	if (pmu->cpuhp_state)
+		cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+
+	ida_simple_remove(&ddr_ida, pmu->id);
+	dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
+	return ret;
+}
+
+static int ddr_perf_remove(struct platform_device *pdev)
+{
+	struct ddr_pmu *pmu = platform_get_drvdata(pdev);
+
+	cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+	irq_set_affinity_hint(pmu->irq, NULL);
+
+	perf_pmu_unregister(&pmu->pmu);
+
+	ida_simple_remove(&ddr_ida, pmu->id);
+	return 0;
+}
+
+static struct platform_driver imx_ddr_pmu_driver = {
+	.driver         = {
+		.name   = "imx-ddr-pmu",
+		.of_match_table = imx_ddr_pmu_dt_ids,
+	},
+	.probe          = ddr_perf_probe,
+	.remove         = ddr_perf_remove,
+};
+
+module_platform_driver(imx_ddr_pmu_driver);
+MODULE_LICENSE("GPL v2");
-- 
2.5.2

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-05-01 18:43   ` Frank Li
  0 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add ddr performance monitor support for iMX8QXP

There are 4 counters for ddr perfomance events.
counter 0 is dedicated for cycles.
you choose any up to 3 no cycles events.

for example:

perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls

Support below events.

  imx8_ddr0/activate/                                     [Kernel PMU event]
  imx8_ddr0/axid-read/                                    [Kernel PMU event]
  imx8_ddr0/axid-write/                                   [Kernel PMU event]
  imx8_ddr0/cycles/                                       [Kernel PMU event]
  imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
  imx8_ddr0/hp-read/                                      [Kernel PMU event]
  imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
  imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/load-mode/                                    [Kernel PMU event]
  imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
  imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
  imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/mwr/                                          [Kernel PMU event]
  imx8_ddr0/precharge/                                    [Kernel PMU event]
  imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
  imx8_ddr0/read-access/                                  [Kernel PMU event]
  imx8_ddr0/read-activate/                                [Kernel PMU event]
  imx8_ddr0/read-command/                                 [Kernel PMU event]
  imx8_ddr0/read-cycles/                                  [Kernel PMU event]
  imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
  imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
  imx8_ddr0/read-write-transition/                        [Kernel PMU event]
  imx8_ddr0/read/                                         [Kernel PMU event]
  imx8_ddr0/refresh/                                      [Kernel PMU event]
  imx8_ddr0/selfresh/                                     [Kernel PMU event]
  imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
  imx8_ddr0/write-access/                                 [Kernel PMU event]
  imx8_ddr0/write-command/                                [Kernel PMU event]
  imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
  imx8_ddr0/write-cycles/                                 [Kernel PMU event]
  imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
  imx8_ddr0/write/

Signed-off-by: Frank Li <Frank.Li@nxp.com>
---

Notes:
    Change from v11 to v12
    * change name from imx_ddr to imx8_ddr

    Change from v10 to v11
     * using devm_platform_ioremap_resource
     * using help macro IMX8_DDR_PMU_EVENT_ATTR

    Change from v9 to v10
     * use name imx8_ddr_perf_pmu
     * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
     * config: 0-7
     * use read_relaxed() at ddr_perf_read_counter
     * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
     * raname ddr_perf_event_start to ddr_perf_counter_enable
     * use FIELD_PREP
     * write 0 to disalbe counter at ddr_perf_event_enable
     * update hwc->state
     * remove local64_set at ddr_perf_event_add
     * move perf_pmu_register after irq_register to simplfy err handle

    No change from v8 to v9

    Change from v7 to v8
     * remove unused define
     * change total_events to active_events, change active_events to events
     * remove flags,
     * fix multi line comments code sytle
     * add pmu_enable\disable function
     * disable event at irq handle
     * remove counter check at ddr_perf_free_counter
     * remove pmu->irq check
     * add group check

    Change from v6 to v7
     * added irq affinity handle, ref arm-ccn.c
     * added IRQF_NOBALANCING | IRQF_NO_THREAD
     * added ida_simple_remove at failure path

    Change from v5 to v6
     * fix insmod\rmmod problem
     * remove randunt register read at irq handle
     * change u32 irq to int
     * devm_request_irq use default flags.

    Change from v4 to v5
     * Remove AXI ID filter function

    Change from v3 to v4
     * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
     * sort include
     * remove struct fsl_ddr_devtype_data
     * Added comment need disable control first
     * Added comment about must enable cycle counter
     * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
     * Added comment about cycle counter is fastest one

    Change from v2 to v3
     * remove kfree

    Change from V1 to V2
     * update Kconfig by use i.MX8 instead of i.MX8 QXP
     * remove gpl statememnt since SPDX tag
     * use dev_kzalloc
     * use dev_err
     * commit message show axi_read 0x41\axi_write 0x42
     * commit message show cycles must be enabled
     * Irq only issue at cycles overflow
     * use NUM_COUNTER
     * use devm_request_irq
     * add hotplug callback to handle context migration

 drivers/perf/Kconfig             |   7 +
 drivers/perf/Makefile            |   1 +
 drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 560 insertions(+)
 create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c

diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index a94e586..9bc3785 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -70,6 +70,13 @@ config ARM_DSU_PMU
 	  system, control logic. The PMU allows counting various events related
 	  to DSU.
 
+config FSL_IMX8_DDR_PMU
+	tristate "Freescale i.MX8 DDR perf monitor"
+	depends on ARCH_MXC
+	  help
+	  Provides support for ddr perfomance monitor in i.MX8. Provide memory
+	  througput information.
+
 config HISI_PMU
        bool "HiSilicon SoC PMU"
        depends on ARM64 && ACPI
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index 3048994..2ebb4de 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
 obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
 obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
 obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
+obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
 obj-$(CONFIG_HISI_PMU) += hisilicon/
 obj-$(CONFIG_QCOM_L2_PMU)	+= qcom_l2_pmu.o
 obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
new file mode 100644
index 0000000..15b0772
--- /dev/null
+++ b/drivers/perf/fsl_imx8_ddr_perf.c
@@ -0,0 +1,552 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2017 NXP
+ * Copyright 2016 Freescale Semiconductor, Inc.
+ */
+
+#include <linux/bitfield.h>
+#include <linux/init.h>
+#include <linux/interrupt.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/of_irq.h>
+#include <linux/perf_event.h>
+#include <linux/slab.h>
+
+#define COUNTER_CNTL		0x0
+#define COUNTER_READ		0x20
+
+#define COUNTER_DPCR1		0x30
+
+#define CNTL_OVER		0x1
+#define CNTL_CLEAR		0x2
+#define CNTL_EN			0x4
+#define CNTL_EN_MASK		0xFFFFFFFB
+#define CNTL_CLEAR_MASK		0xFFFFFFFD
+#define CNTL_OVER_MASK		0xFFFFFFFE
+
+#define CNTL_CSV_SHIFT		24
+#define CNTL_CSV_MASK		(0xFF << CNTL_CSV_SHIFT)
+
+#define EVENT_CYCLES_ID		0
+#define EVENT_CYCLES_COUNTER	0
+#define NUM_COUNTERS		4
+
+#define to_ddr_pmu(p)		container_of(p, struct ddr_pmu, pmu)
+
+#define DDR_PERF_DEV_NAME	"imx8_ddr_perf_pmu"
+
+static DEFINE_IDA(ddr_ida);
+
+static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
+	{ .compatible = "fsl,imx8-ddr-pmu",},
+	{ .compatible = "fsl,imx8m-ddr-pmu",},
+	{ /* sentinel */ }
+};
+
+struct ddr_pmu {
+	struct pmu pmu;
+	void __iomem *base;
+	unsigned int cpu;
+	struct	hlist_node node;
+	struct	device *dev;
+	struct perf_event *events[NUM_COUNTERS];
+	int active_events;
+	enum cpuhp_state cpuhp_state;
+	int irq;
+	int id;
+};
+
+static ssize_t ddr_perf_cpumask_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct ddr_pmu *pmu = dev_get_drvdata(dev);
+
+	return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
+}
+
+static struct device_attribute ddr_perf_cpumask_attr =
+	__ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
+
+static struct attribute *ddr_perf_cpumask_attrs[] = {
+	&ddr_perf_cpumask_attr.attr,
+	NULL,
+};
+
+static struct attribute_group ddr_perf_cpumask_attr_group = {
+	.attrs = ddr_perf_cpumask_attrs,
+};
+
+static ssize_t
+ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)				\
+	(&((struct perf_pmu_events_attr[]) {				\
+		{ .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
+		  .id = _id, }						\
+	})[0].attr.attr)
+
+static struct attribute *ddr_perf_events_attrs[] = {
+	IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
+	IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
+	IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
+	IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
+	IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
+	IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
+	IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
+	IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
+	IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
+	IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
+	IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
+	IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
+	IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
+	IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
+	IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
+	IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
+	IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
+	IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
+	IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
+	IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
+	IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
+	IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
+	IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
+	IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
+	IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
+	NULL,
+};
+
+static struct attribute_group ddr_perf_events_attr_group = {
+	.name = "events",
+	.attrs = ddr_perf_events_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-7");
+
+static struct attribute *ddr_perf_format_attrs[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group ddr_perf_format_attr_group = {
+	.name = "format",
+	.attrs = ddr_perf_format_attrs,
+};
+
+static const struct attribute_group *attr_groups[] = {
+	&ddr_perf_events_attr_group,
+	&ddr_perf_format_attr_group,
+	&ddr_perf_cpumask_attr_group,
+	NULL,
+};
+
+static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
+{
+	int i;
+
+	/*
+	 * Always map cycle event to counter 0
+	 * Cycles counter is dedicated for cycle event
+	 * can't used for the other events
+	 */
+	if (event == EVENT_CYCLES_ID) {
+		if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+			return EVENT_CYCLES_COUNTER;
+		else
+			return -ENOENT;
+	}
+
+	for (i = 1; i < NUM_COUNTERS; i++) {
+		if (pmu->events[i] == NULL)
+			return i;
+	}
+
+	return -ENOENT;
+}
+
+static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
+{
+	pmu->events[counter] = NULL;
+}
+
+static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
+{
+	return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
+}
+
+static int ddr_perf_event_init(struct perf_event *event)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	struct perf_event *sibling;
+
+	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) {
+		dev_warn(pmu->dev, "Can't provide per-task data!\n");
+		return -EOPNOTSUPP;
+	}
+
+	/*
+	 * We must NOT create groups containing mixed PMUs, although software
+	 * events are acceptable (for example to create a CCN group
+	 * periodically read when a hrtimer aka cpu-clock leader triggers).
+	 */
+	if (event->group_leader->pmu != event->pmu &&
+			!is_software_event(event->group_leader))
+		return -EINVAL;
+
+	for_each_sibling_event(sibling, event->group_leader) {
+		if (sibling->pmu != event->pmu &&
+				!is_software_event(sibling))
+			return -EINVAL;
+	}
+
+	event->cpu = pmu->cpu;
+	hwc->idx = -1;
+
+	return 0;
+}
+
+
+static void ddr_perf_event_update(struct perf_event *event)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	u64 delta, prev_raw_count, new_raw_count;
+	int counter = hwc->idx;
+
+	do {
+		prev_raw_count = local64_read(&hwc->prev_count);
+		new_raw_count = ddr_perf_read_counter(pmu, counter);
+	} while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+			new_raw_count) != prev_raw_count);
+
+	delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
+
+	local64_add(delta, &event->count);
+}
+
+static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
+				  int counter, bool enable)
+{
+	u8 reg = counter * 4 + COUNTER_CNTL;
+	int val;
+
+	if (enable) {
+		/*
+		 * must disable first, then enable again
+		 * otherwise, cycle counter will not work
+		 * if previous state is enabled.
+		 */
+		writel(0, pmu->base + reg);
+		val = CNTL_EN | CNTL_CLEAR;
+		val |= FIELD_PREP(CNTL_CSV_MASK, config);
+		writel(val, pmu->base + reg);
+	} else {
+		/* Disable counter */
+		writel(0, pmu->base + reg);
+	}
+}
+
+static void ddr_perf_event_start(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	local64_set(&hwc->prev_count, 0);
+
+	ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
+
+	hwc->state = 0;
+}
+
+static int ddr_perf_event_add(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter;
+	int cfg = event->attr.config;
+
+	counter = ddr_perf_alloc_counter(pmu, cfg);
+	if (counter < 0) {
+		dev_dbg(pmu->dev, "There are not enough counters\n");
+		return -EOPNOTSUPP;
+	}
+
+	pmu->events[counter] = event;
+	pmu->active_events++;
+	hwc->idx = counter;
+
+	hwc->state |= PERF_HES_STOPPED;
+
+	if (flags & PERF_EF_START)
+		ddr_perf_event_start(event, flags);
+
+	return 0;
+}
+
+static void ddr_perf_event_stop(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
+	ddr_perf_event_update(event);
+
+	hwc->state |= PERF_HES_STOPPED;
+}
+
+static void ddr_perf_event_del(struct perf_event *event, int flags)
+{
+	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	int counter = hwc->idx;
+
+	ddr_perf_event_stop(event, PERF_EF_UPDATE);
+
+	ddr_perf_free_counter(pmu, counter);
+	pmu->active_events--;
+	hwc->idx = -1;
+}
+
+static void ddr_perf_pmu_enable(struct pmu *pmu)
+{
+	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
+
+	/* enable cycle counter if cycle is not active event list */
+	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+		ddr_perf_counter_enable(ddr_pmu,
+				      EVENT_CYCLES_ID,
+				      EVENT_CYCLES_COUNTER,
+				      true);
+}
+
+static void ddr_perf_pmu_disable(struct pmu *pmu)
+{
+	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
+
+	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
+		ddr_perf_counter_enable(ddr_pmu,
+				      EVENT_CYCLES_ID,
+				      EVENT_CYCLES_COUNTER,
+				      false);
+}
+
+static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
+			 struct device *dev)
+{
+	*pmu = (struct ddr_pmu) {
+		.pmu = (struct pmu) {
+			.capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+			.task_ctx_nr = perf_invalid_context,
+			.attr_groups = attr_groups,
+			.event_init  = ddr_perf_event_init,
+			.add	     = ddr_perf_event_add,
+			.del	     = ddr_perf_event_del,
+			.start	     = ddr_perf_event_start,
+			.stop	     = ddr_perf_event_stop,
+			.read	     = ddr_perf_event_update,
+			.pmu_enable  = ddr_perf_pmu_enable,
+			.pmu_disable = ddr_perf_pmu_disable,
+		},
+		.base = base,
+		.dev = dev,
+	};
+
+	pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
+	return pmu->id;
+}
+
+static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
+{
+	int i;
+	struct ddr_pmu *pmu = (struct ddr_pmu *) p;
+	struct perf_event *event, *cycle_event = NULL;
+
+	/* all counter will stop if cycle counter disabled */
+	ddr_perf_counter_enable(pmu,
+			      EVENT_CYCLES_ID,
+			      EVENT_CYCLES_COUNTER,
+			      false);
+	/*
+	 * When the cycle counter overflows, all counters are stopped,
+	 * and an IRQ is raised. If any other counter overflows, it
+	 * continues counting, and no IRQ is raised.
+	 *
+	 * Cycles occur at least 4 times as often as other events, so we
+	 * can update all events on a cycle counter overflow and not
+	 * lose events.
+	 *
+	 */
+	for (i = 0; i < NUM_COUNTERS; i++) {
+
+		if (!pmu->events[i])
+			continue;
+
+		event = pmu->events[i];
+
+		ddr_perf_event_update(event);
+
+		if (event->hw.idx == EVENT_CYCLES_COUNTER)
+			cycle_event = event;
+	}
+
+	ddr_perf_counter_enable(pmu,
+			      EVENT_CYCLES_ID,
+			      EVENT_CYCLES_COUNTER,
+			      true);
+	if (cycle_event)
+		ddr_perf_event_update(cycle_event);
+
+	return IRQ_HANDLED;
+}
+
+static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+	struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
+	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;
+
+	WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
+
+	return 0;
+}
+
+static int ddr_perf_probe(struct platform_device *pdev)
+{
+	struct ddr_pmu *pmu;
+	struct device_node *np;
+	void __iomem *base;
+	char *name;
+	int num;
+	int ret;
+	int irq;
+
+	base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	np = pdev->dev.of_node;
+
+	pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
+	if (!pmu)
+		return -ENOMEM;
+
+	num = ddr_perf_init(pmu, base, &pdev->dev);
+
+	platform_set_drvdata(pdev, pmu);
+
+	name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
+	if (!name)
+		return -ENOMEM;
+
+	pmu->cpu = raw_smp_processor_id();
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
+				      DDR_PERF_DEV_NAME,
+				      NULL,
+				      ddr_perf_offline_cpu);
+
+	if (ret < 0) {
+		dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
+		goto ddr_perf_err;
+	}
+
+	pmu->cpuhp_state = ret;
+
+	/* Register the pmu instance for cpu hotplug */
+	cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+
+	/* Request irq */
+	irq = of_irq_get(np, 0);
+	if (irq < 0) {
+		dev_err(&pdev->dev, "Failed to get irq: %d", irq);
+		ret = irq;
+		goto ddr_perf_err;
+	}
+
+	ret = devm_request_irq(&pdev->dev, irq,
+					ddr_perf_irq_handler,
+					IRQF_NOBALANCING | IRQF_NO_THREAD,
+					DDR_PERF_DEV_NAME,
+					pmu);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "Request irq failed: %d", ret);
+		goto ddr_perf_err;
+	}
+
+	pmu->irq = irq;
+	ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
+	if (ret) {
+		dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
+		goto ddr_perf_err;
+	}
+
+	ret = perf_pmu_register(&pmu->pmu, name, -1);
+	if (ret)
+		goto ddr_perf_err;
+
+	return 0;
+
+ddr_perf_err:
+	if (pmu->cpuhp_state)
+		cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+
+	ida_simple_remove(&ddr_ida, pmu->id);
+	dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
+	return ret;
+}
+
+static int ddr_perf_remove(struct platform_device *pdev)
+{
+	struct ddr_pmu *pmu = platform_get_drvdata(pdev);
+
+	cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
+	irq_set_affinity_hint(pmu->irq, NULL);
+
+	perf_pmu_unregister(&pmu->pmu);
+
+	ida_simple_remove(&ddr_ida, pmu->id);
+	return 0;
+}
+
+static struct platform_driver imx_ddr_pmu_driver = {
+	.driver         = {
+		.name   = "imx-ddr-pmu",
+		.of_match_table = imx_ddr_pmu_dt_ids,
+	},
+	.probe          = ddr_perf_probe,
+	.remove         = ddr_perf_remove,
+};
+
+module_platform_driver(imx_ddr_pmu_driver);
+MODULE_LICENSE("GPL v2");
-- 
2.5.2

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 3/4] arm64: dts: imx8qxp: added ddr performance monitor nodes
  2019-05-01 18:43 ` Frank Li
@ 2019-05-01 18:43   ` Frank Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add ddr performance monitor

Signed-off-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
---

Notes:
    No change from v9 to v12

    Change from v8 to v9
    * put ddr-pmu under ddr_subsystem bus

    Change from v3 to v8
    * none

    Change from v2 to v3
    * ddr_pmu0 -> ddr-pmu

 arch/arm64/boot/dts/freescale/imx8qxp.dtsi | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
index 0683ee2..a33e08c 100644
--- a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
@@ -378,6 +378,20 @@
 		};
 	};
 
+	ddr_subsyss: bus@5c000000 {
+		compatible = "simple-bus";
+		#address-cells = <1>;
+		#size-cells = <1>;
+		ranges = <0x5c000000 0x0 0x5c000000 0x1000000>;
+
+		ddr-pmu@5c020000 {
+			compatible = "fsl,imx8-ddr-pmu";
+			reg = <0x5c020000 0x10000>;
+			interrupt-parent = <&gic>;
+			interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+		};
+	};
+
 	lsio_subsys: bus@5d000000 {
 		compatible = "simple-bus";
 		#address-cells = <1>;
-- 
2.5.2

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 3/4] arm64: dts: imx8qxp: added ddr performance monitor nodes
@ 2019-05-01 18:43   ` Frank Li
  0 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add ddr performance monitor

Signed-off-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
---

Notes:
    No change from v9 to v12

    Change from v8 to v9
    * put ddr-pmu under ddr_subsystem bus

    Change from v3 to v8
    * none

    Change from v2 to v3
    * ddr_pmu0 -> ddr-pmu

 arch/arm64/boot/dts/freescale/imx8qxp.dtsi | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
index 0683ee2..a33e08c 100644
--- a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
@@ -378,6 +378,20 @@
 		};
 	};
 
+	ddr_subsyss: bus@5c000000 {
+		compatible = "simple-bus";
+		#address-cells = <1>;
+		#size-cells = <1>;
+		ranges = <0x5c000000 0x0 0x5c000000 0x1000000>;
+
+		ddr-pmu@5c020000 {
+			compatible = "fsl,imx8-ddr-pmu";
+			reg = <0x5c020000 0x10000>;
+			interrupt-parent = <&gic>;
+			interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+		};
+	};
+
 	lsio_subsys: bus@5d000000 {
 		compatible = "simple-bus";
 		#address-cells = <1>;
-- 
2.5.2

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 4/4] MAINTAINERS: Added imx DDR performonitor driver maintainer information
  2019-05-01 18:43 ` Frank Li
@ 2019-05-01 18:43   ` Frank Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add DDR perf counter driver maintainer information

Signed-off-by: Frank Li <Frank.Li@nxp.com>
---

Notes:
    No change from v1 to v12

 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 2c7d4e1..6a9868b 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -6264,6 +6264,13 @@ L:	linux-i2c@vger.kernel.org
 S:	Maintained
 F:	drivers/i2c/busses/i2c-cpm.c
 
+FREESCALE IMX DDR Performance Monitor DRIVER
+M:	Frank Li <Frank.li@nxp.com>
+L:	linux-arm-kernel@lists.infradead.org
+S:	Maintained
+F:	drivers/perf/fsl_imx8_ddr_perf.c
+F:	Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
+
 FREESCALE IMX LPI2C DRIVER
 M:	Dong Aisheng <aisheng.dong@nxp.com>
 L:	linux-i2c@vger.kernel.org
-- 
2.5.2

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* [PATCH V12 4/4] MAINTAINERS: Added imx DDR performonitor driver maintainer information
@ 2019-05-01 18:43   ` Frank Li
  0 siblings, 0 replies; 30+ messages in thread
From: Frank Li @ 2019-05-01 18:43 UTC (permalink / raw)
  To: andrew.smirnov, mark.rutland, will.deacon, shawnguo, s.hauer,
	kernel, festevam, dl-linux-imx, robh+dt, Aisheng Dong,
	devicetree, lznuaa, linux-arm-kernel
  Cc: Frank Li

Add DDR perf counter driver maintainer information

Signed-off-by: Frank Li <Frank.Li@nxp.com>
---

Notes:
    No change from v1 to v12

 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 2c7d4e1..6a9868b 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -6264,6 +6264,13 @@ L:	linux-i2c@vger.kernel.org
 S:	Maintained
 F:	drivers/i2c/busses/i2c-cpm.c
 
+FREESCALE IMX DDR Performance Monitor DRIVER
+M:	Frank Li <Frank.li@nxp.com>
+L:	linux-arm-kernel@lists.infradead.org
+S:	Maintained
+F:	drivers/perf/fsl_imx8_ddr_perf.c
+F:	Documentation/devicetree/bindings/perf/fsl-imx-ddr.txt
+
 FREESCALE IMX LPI2C DRIVER
 M:	Dong Aisheng <aisheng.dong@nxp.com>
 L:	linux-i2c@vger.kernel.org
-- 
2.5.2

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply related	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-05-01 18:43   ` Frank Li
@ 2019-05-07 19:30     ` Andrey Smirnov
  -1 siblings, 0 replies; 30+ messages in thread
From: Andrey Smirnov @ 2019-05-07 19:30 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, festevam, s.hauer,
	will.deacon, robh+dt, dl-linux-imx, kernel, lznuaa, shawnguo,
	linux-arm-kernel

On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
>
> Add ddr performance monitor support for iMX8QXP
>
> There are 4 counters for ddr perfomance events.
> counter 0 is dedicated for cycles.
> you choose any up to 3 no cycles events.
>
> for example:
>
> perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
>
> Support below events.
>
>   imx8_ddr0/activate/                                     [Kernel PMU event]
>   imx8_ddr0/axid-read/                                    [Kernel PMU event]
>   imx8_ddr0/axid-write/                                   [Kernel PMU event]
>   imx8_ddr0/cycles/                                       [Kernel PMU event]
>   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
>   imx8_ddr0/hp-read/                                      [Kernel PMU event]
>   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
>   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/load-mode/                                    [Kernel PMU event]
>   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
>   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
>   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/mwr/                                          [Kernel PMU event]
>   imx8_ddr0/precharge/                                    [Kernel PMU event]
>   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
>   imx8_ddr0/read-access/                                  [Kernel PMU event]
>   imx8_ddr0/read-activate/                                [Kernel PMU event]
>   imx8_ddr0/read-command/                                 [Kernel PMU event]
>   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
>   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
>   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
>   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
>   imx8_ddr0/read/                                         [Kernel PMU event]
>   imx8_ddr0/refresh/                                      [Kernel PMU event]
>   imx8_ddr0/selfresh/                                     [Kernel PMU event]
>   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/write-access/                                 [Kernel PMU event]
>   imx8_ddr0/write-command/                                [Kernel PMU event]
>   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
>   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
>   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
>   imx8_ddr0/write/
>
> Signed-off-by: Frank Li <Frank.Li@nxp.com>

Don't really have anything to suggest, that I already haven't. LGTM, so:

Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>


> ---
>
> Notes:
>     Change from v11 to v12
>     * change name from imx_ddr to imx8_ddr
>
>     Change from v10 to v11
>      * using devm_platform_ioremap_resource
>      * using help macro IMX8_DDR_PMU_EVENT_ATTR
>
>     Change from v9 to v10
>      * use name imx8_ddr_perf_pmu
>      * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
>      * config: 0-7
>      * use read_relaxed() at ddr_perf_read_counter
>      * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
>      * raname ddr_perf_event_start to ddr_perf_counter_enable
>      * use FIELD_PREP
>      * write 0 to disalbe counter at ddr_perf_event_enable
>      * update hwc->state
>      * remove local64_set at ddr_perf_event_add
>      * move perf_pmu_register after irq_register to simplfy err handle
>
>     No change from v8 to v9
>
>     Change from v7 to v8
>      * remove unused define
>      * change total_events to active_events, change active_events to events
>      * remove flags,
>      * fix multi line comments code sytle
>      * add pmu_enable\disable function
>      * disable event at irq handle
>      * remove counter check at ddr_perf_free_counter
>      * remove pmu->irq check
>      * add group check
>
>     Change from v6 to v7
>      * added irq affinity handle, ref arm-ccn.c
>      * added IRQF_NOBALANCING | IRQF_NO_THREAD
>      * added ida_simple_remove at failure path
>
>     Change from v5 to v6
>      * fix insmod\rmmod problem
>      * remove randunt register read at irq handle
>      * change u32 irq to int
>      * devm_request_irq use default flags.
>
>     Change from v4 to v5
>      * Remove AXI ID filter function
>
>     Change from v3 to v4
>      * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
>      * sort include
>      * remove struct fsl_ddr_devtype_data
>      * Added comment need disable control first
>      * Added comment about must enable cycle counter
>      * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
>      * Added comment about cycle counter is fastest one
>
>     Change from v2 to v3
>      * remove kfree
>
>     Change from V1 to V2
>      * update Kconfig by use i.MX8 instead of i.MX8 QXP
>      * remove gpl statememnt since SPDX tag
>      * use dev_kzalloc
>      * use dev_err
>      * commit message show axi_read 0x41\axi_write 0x42
>      * commit message show cycles must be enabled
>      * Irq only issue at cycles overflow
>      * use NUM_COUNTER
>      * use devm_request_irq
>      * add hotplug callback to handle context migration
>
>  drivers/perf/Kconfig             |   7 +
>  drivers/perf/Makefile            |   1 +
>  drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
>  3 files changed, 560 insertions(+)
>  create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c
>
> diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
> index a94e586..9bc3785 100644
> --- a/drivers/perf/Kconfig
> +++ b/drivers/perf/Kconfig
> @@ -70,6 +70,13 @@ config ARM_DSU_PMU
>           system, control logic. The PMU allows counting various events related
>           to DSU.
>
> +config FSL_IMX8_DDR_PMU
> +       tristate "Freescale i.MX8 DDR perf monitor"
> +       depends on ARCH_MXC
> +         help
> +         Provides support for ddr perfomance monitor in i.MX8. Provide memory
> +         througput information.
> +
>  config HISI_PMU
>         bool "HiSilicon SoC PMU"
>         depends on ARM64 && ACPI
> diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
> index 3048994..2ebb4de 100644
> --- a/drivers/perf/Makefile
> +++ b/drivers/perf/Makefile
> @@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
>  obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
>  obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
>  obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
> +obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
>  obj-$(CONFIG_HISI_PMU) += hisilicon/
>  obj-$(CONFIG_QCOM_L2_PMU)      += qcom_l2_pmu.o
>  obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
> diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
> new file mode 100644
> index 0000000..15b0772
> --- /dev/null
> +++ b/drivers/perf/fsl_imx8_ddr_perf.c
> @@ -0,0 +1,552 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright 2017 NXP
> + * Copyright 2016 Freescale Semiconductor, Inc.
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/init.h>
> +#include <linux/interrupt.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/of_irq.h>
> +#include <linux/perf_event.h>
> +#include <linux/slab.h>
> +
> +#define COUNTER_CNTL           0x0
> +#define COUNTER_READ           0x20
> +
> +#define COUNTER_DPCR1          0x30
> +
> +#define CNTL_OVER              0x1
> +#define CNTL_CLEAR             0x2
> +#define CNTL_EN                        0x4
> +#define CNTL_EN_MASK           0xFFFFFFFB
> +#define CNTL_CLEAR_MASK                0xFFFFFFFD
> +#define CNTL_OVER_MASK         0xFFFFFFFE
> +
> +#define CNTL_CSV_SHIFT         24
> +#define CNTL_CSV_MASK          (0xFF << CNTL_CSV_SHIFT)
> +
> +#define EVENT_CYCLES_ID                0
> +#define EVENT_CYCLES_COUNTER   0
> +#define NUM_COUNTERS           4
> +
> +#define to_ddr_pmu(p)          container_of(p, struct ddr_pmu, pmu)
> +
> +#define DDR_PERF_DEV_NAME      "imx8_ddr_perf_pmu"
> +
> +static DEFINE_IDA(ddr_ida);
> +
> +static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
> +       { .compatible = "fsl,imx8-ddr-pmu",},
> +       { .compatible = "fsl,imx8m-ddr-pmu",},
> +       { /* sentinel */ }
> +};
> +
> +struct ddr_pmu {
> +       struct pmu pmu;
> +       void __iomem *base;
> +       unsigned int cpu;
> +       struct  hlist_node node;
> +       struct  device *dev;
> +       struct perf_event *events[NUM_COUNTERS];
> +       int active_events;
> +       enum cpuhp_state cpuhp_state;
> +       int irq;
> +       int id;
> +};
> +
> +static ssize_t ddr_perf_cpumask_show(struct device *dev,
> +                               struct device_attribute *attr, char *buf)
> +{
> +       struct ddr_pmu *pmu = dev_get_drvdata(dev);
> +
> +       return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
> +}
> +
> +static struct device_attribute ddr_perf_cpumask_attr =
> +       __ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
> +
> +static struct attribute *ddr_perf_cpumask_attrs[] = {
> +       &ddr_perf_cpumask_attr.attr,
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_cpumask_attr_group = {
> +       .attrs = ddr_perf_cpumask_attrs,
> +};
> +
> +static ssize_t
> +ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)                            \
> +       (&((struct perf_pmu_events_attr[]) {                            \
> +               { .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
> +                 .id = _id, }                                          \
> +       })[0].attr.attr)
> +
> +static struct attribute *ddr_perf_events_attrs[] = {
> +       IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
> +       IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
> +       IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
> +       IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
> +       IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
> +       IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
> +       IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
> +       IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
> +       IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
> +       IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
> +       IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_events_attr_group = {
> +       .name = "events",
> +       .attrs = ddr_perf_events_attrs,
> +};
> +
> +PMU_FORMAT_ATTR(event, "config:0-7");
> +
> +static struct attribute *ddr_perf_format_attrs[] = {
> +       &format_attr_event.attr,
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_format_attr_group = {
> +       .name = "format",
> +       .attrs = ddr_perf_format_attrs,
> +};
> +
> +static const struct attribute_group *attr_groups[] = {
> +       &ddr_perf_events_attr_group,
> +       &ddr_perf_format_attr_group,
> +       &ddr_perf_cpumask_attr_group,
> +       NULL,
> +};
> +
> +static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
> +{
> +       int i;
> +
> +       /*
> +        * Always map cycle event to counter 0
> +        * Cycles counter is dedicated for cycle event
> +        * can't used for the other events
> +        */
> +       if (event == EVENT_CYCLES_ID) {
> +               if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +                       return EVENT_CYCLES_COUNTER;
> +               else
> +                       return -ENOENT;
> +       }
> +
> +       for (i = 1; i < NUM_COUNTERS; i++) {
> +               if (pmu->events[i] == NULL)
> +                       return i;
> +       }
> +
> +       return -ENOENT;
> +}
> +
> +static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
> +{
> +       pmu->events[counter] = NULL;
> +}
> +
> +static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
> +{
> +       return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
> +}
> +
> +static int ddr_perf_event_init(struct perf_event *event)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       struct perf_event *sibling;
> +
> +       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) {
> +               dev_warn(pmu->dev, "Can't provide per-task data!\n");
> +               return -EOPNOTSUPP;
> +       }
> +
> +       /*
> +        * We must NOT create groups containing mixed PMUs, although software
> +        * events are acceptable (for example to create a CCN group
> +        * periodically read when a hrtimer aka cpu-clock leader triggers).
> +        */
> +       if (event->group_leader->pmu != event->pmu &&
> +                       !is_software_event(event->group_leader))
> +               return -EINVAL;
> +
> +       for_each_sibling_event(sibling, event->group_leader) {
> +               if (sibling->pmu != event->pmu &&
> +                               !is_software_event(sibling))
> +                       return -EINVAL;
> +       }
> +
> +       event->cpu = pmu->cpu;
> +       hwc->idx = -1;
> +
> +       return 0;
> +}
> +
> +
> +static void ddr_perf_event_update(struct perf_event *event)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       u64 delta, prev_raw_count, new_raw_count;
> +       int counter = hwc->idx;
> +
> +       do {
> +               prev_raw_count = local64_read(&hwc->prev_count);
> +               new_raw_count = ddr_perf_read_counter(pmu, counter);
> +       } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
> +                       new_raw_count) != prev_raw_count);
> +
> +       delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
> +
> +       local64_add(delta, &event->count);
> +}
> +
> +static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
> +                                 int counter, bool enable)
> +{
> +       u8 reg = counter * 4 + COUNTER_CNTL;
> +       int val;
> +
> +       if (enable) {
> +               /*
> +                * must disable first, then enable again
> +                * otherwise, cycle counter will not work
> +                * if previous state is enabled.
> +                */
> +               writel(0, pmu->base + reg);
> +               val = CNTL_EN | CNTL_CLEAR;
> +               val |= FIELD_PREP(CNTL_CSV_MASK, config);
> +               writel(val, pmu->base + reg);
> +       } else {
> +               /* Disable counter */
> +               writel(0, pmu->base + reg);
> +       }
> +}
> +
> +static void ddr_perf_event_start(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       local64_set(&hwc->prev_count, 0);
> +
> +       ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
> +
> +       hwc->state = 0;
> +}
> +
> +static int ddr_perf_event_add(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter;
> +       int cfg = event->attr.config;
> +
> +       counter = ddr_perf_alloc_counter(pmu, cfg);
> +       if (counter < 0) {
> +               dev_dbg(pmu->dev, "There are not enough counters\n");
> +               return -EOPNOTSUPP;
> +       }
> +
> +       pmu->events[counter] = event;
> +       pmu->active_events++;
> +       hwc->idx = counter;
> +
> +       hwc->state |= PERF_HES_STOPPED;
> +
> +       if (flags & PERF_EF_START)
> +               ddr_perf_event_start(event, flags);
> +
> +       return 0;
> +}
> +
> +static void ddr_perf_event_stop(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
> +       ddr_perf_event_update(event);
> +
> +       hwc->state |= PERF_HES_STOPPED;
> +}
> +
> +static void ddr_perf_event_del(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       ddr_perf_event_stop(event, PERF_EF_UPDATE);
> +
> +       ddr_perf_free_counter(pmu, counter);
> +       pmu->active_events--;
> +       hwc->idx = -1;
> +}
> +
> +static void ddr_perf_pmu_enable(struct pmu *pmu)
> +{
> +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> +
> +       /* enable cycle counter if cycle is not active event list */
> +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +               ddr_perf_counter_enable(ddr_pmu,
> +                                     EVENT_CYCLES_ID,
> +                                     EVENT_CYCLES_COUNTER,
> +                                     true);
> +}
> +
> +static void ddr_perf_pmu_disable(struct pmu *pmu)
> +{
> +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> +
> +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +               ddr_perf_counter_enable(ddr_pmu,
> +                                     EVENT_CYCLES_ID,
> +                                     EVENT_CYCLES_COUNTER,
> +                                     false);
> +}
> +
> +static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
> +                        struct device *dev)
> +{
> +       *pmu = (struct ddr_pmu) {
> +               .pmu = (struct pmu) {
> +                       .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
> +                       .task_ctx_nr = perf_invalid_context,
> +                       .attr_groups = attr_groups,
> +                       .event_init  = ddr_perf_event_init,
> +                       .add         = ddr_perf_event_add,
> +                       .del         = ddr_perf_event_del,
> +                       .start       = ddr_perf_event_start,
> +                       .stop        = ddr_perf_event_stop,
> +                       .read        = ddr_perf_event_update,
> +                       .pmu_enable  = ddr_perf_pmu_enable,
> +                       .pmu_disable = ddr_perf_pmu_disable,
> +               },
> +               .base = base,
> +               .dev = dev,
> +       };
> +
> +       pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
> +       return pmu->id;
> +}
> +
> +static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
> +{
> +       int i;
> +       struct ddr_pmu *pmu = (struct ddr_pmu *) p;
> +       struct perf_event *event, *cycle_event = NULL;
> +
> +       /* all counter will stop if cycle counter disabled */
> +       ddr_perf_counter_enable(pmu,
> +                             EVENT_CYCLES_ID,
> +                             EVENT_CYCLES_COUNTER,
> +                             false);
> +       /*
> +        * When the cycle counter overflows, all counters are stopped,
> +        * and an IRQ is raised. If any other counter overflows, it
> +        * continues counting, and no IRQ is raised.
> +        *
> +        * Cycles occur at least 4 times as often as other events, so we
> +        * can update all events on a cycle counter overflow and not
> +        * lose events.
> +        *
> +        */
> +       for (i = 0; i < NUM_COUNTERS; i++) {
> +
> +               if (!pmu->events[i])
> +                       continue;
> +
> +               event = pmu->events[i];
> +
> +               ddr_perf_event_update(event);
> +
> +               if (event->hw.idx == EVENT_CYCLES_COUNTER)
> +                       cycle_event = event;
> +       }
> +
> +       ddr_perf_counter_enable(pmu,
> +                             EVENT_CYCLES_ID,
> +                             EVENT_CYCLES_COUNTER,
> +                             true);
> +       if (cycle_event)
> +               ddr_perf_event_update(cycle_event);
> +
> +       return IRQ_HANDLED;
> +}
> +
> +static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
> +{
> +       struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
> +       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;
> +
> +       WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
> +
> +       return 0;
> +}
> +
> +static int ddr_perf_probe(struct platform_device *pdev)
> +{
> +       struct ddr_pmu *pmu;
> +       struct device_node *np;
> +       void __iomem *base;
> +       char *name;
> +       int num;
> +       int ret;
> +       int irq;
> +
> +       base = devm_platform_ioremap_resource(pdev, 0);
> +       if (IS_ERR(base))
> +               return PTR_ERR(base);
> +
> +       np = pdev->dev.of_node;
> +
> +       pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
> +       if (!pmu)
> +               return -ENOMEM;
> +
> +       num = ddr_perf_init(pmu, base, &pdev->dev);
> +
> +       platform_set_drvdata(pdev, pmu);
> +
> +       name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
> +       if (!name)
> +               return -ENOMEM;
> +
> +       pmu->cpu = raw_smp_processor_id();
> +       ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> +                                     DDR_PERF_DEV_NAME,
> +                                     NULL,
> +                                     ddr_perf_offline_cpu);
> +
> +       if (ret < 0) {
> +               dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
> +               goto ddr_perf_err;
> +       }
> +
> +       pmu->cpuhp_state = ret;
> +
> +       /* Register the pmu instance for cpu hotplug */
> +       cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +
> +       /* Request irq */
> +       irq = of_irq_get(np, 0);
> +       if (irq < 0) {
> +               dev_err(&pdev->dev, "Failed to get irq: %d", irq);
> +               ret = irq;
> +               goto ddr_perf_err;
> +       }
> +
> +       ret = devm_request_irq(&pdev->dev, irq,
> +                                       ddr_perf_irq_handler,
> +                                       IRQF_NOBALANCING | IRQF_NO_THREAD,
> +                                       DDR_PERF_DEV_NAME,
> +                                       pmu);
> +       if (ret < 0) {
> +               dev_err(&pdev->dev, "Request irq failed: %d", ret);
> +               goto ddr_perf_err;
> +       }
> +
> +       pmu->irq = irq;
> +       ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
> +       if (ret) {
> +               dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
> +               goto ddr_perf_err;
> +       }
> +
> +       ret = perf_pmu_register(&pmu->pmu, name, -1);
> +       if (ret)
> +               goto ddr_perf_err;
> +
> +       return 0;
> +
> +ddr_perf_err:
> +       if (pmu->cpuhp_state)
> +               cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +
> +       ida_simple_remove(&ddr_ida, pmu->id);
> +       dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
> +       return ret;
> +}
> +
> +static int ddr_perf_remove(struct platform_device *pdev)
> +{
> +       struct ddr_pmu *pmu = platform_get_drvdata(pdev);
> +
> +       cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +       irq_set_affinity_hint(pmu->irq, NULL);
> +
> +       perf_pmu_unregister(&pmu->pmu);
> +
> +       ida_simple_remove(&ddr_ida, pmu->id);
> +       return 0;
> +}
> +
> +static struct platform_driver imx_ddr_pmu_driver = {
> +       .driver         = {
> +               .name   = "imx-ddr-pmu",
> +               .of_match_table = imx_ddr_pmu_dt_ids,
> +       },
> +       .probe          = ddr_perf_probe,
> +       .remove         = ddr_perf_remove,
> +};
> +
> +module_platform_driver(imx_ddr_pmu_driver);
> +MODULE_LICENSE("GPL v2");
> --
> 2.5.2
>

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-05-07 19:30     ` Andrey Smirnov
  0 siblings, 0 replies; 30+ messages in thread
From: Andrey Smirnov @ 2019-05-07 19:30 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, festevam, s.hauer,
	will.deacon, robh+dt, dl-linux-imx, kernel, lznuaa, shawnguo,
	linux-arm-kernel

On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
>
> Add ddr performance monitor support for iMX8QXP
>
> There are 4 counters for ddr perfomance events.
> counter 0 is dedicated for cycles.
> you choose any up to 3 no cycles events.
>
> for example:
>
> perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
>
> Support below events.
>
>   imx8_ddr0/activate/                                     [Kernel PMU event]
>   imx8_ddr0/axid-read/                                    [Kernel PMU event]
>   imx8_ddr0/axid-write/                                   [Kernel PMU event]
>   imx8_ddr0/cycles/                                       [Kernel PMU event]
>   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
>   imx8_ddr0/hp-read/                                      [Kernel PMU event]
>   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
>   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/load-mode/                                    [Kernel PMU event]
>   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
>   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
>   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/mwr/                                          [Kernel PMU event]
>   imx8_ddr0/precharge/                                    [Kernel PMU event]
>   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
>   imx8_ddr0/read-access/                                  [Kernel PMU event]
>   imx8_ddr0/read-activate/                                [Kernel PMU event]
>   imx8_ddr0/read-command/                                 [Kernel PMU event]
>   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
>   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
>   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
>   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
>   imx8_ddr0/read/                                         [Kernel PMU event]
>   imx8_ddr0/refresh/                                      [Kernel PMU event]
>   imx8_ddr0/selfresh/                                     [Kernel PMU event]
>   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
>   imx8_ddr0/write-access/                                 [Kernel PMU event]
>   imx8_ddr0/write-command/                                [Kernel PMU event]
>   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
>   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
>   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
>   imx8_ddr0/write/
>
> Signed-off-by: Frank Li <Frank.Li@nxp.com>

Don't really have anything to suggest, that I already haven't. LGTM, so:

Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>


> ---
>
> Notes:
>     Change from v11 to v12
>     * change name from imx_ddr to imx8_ddr
>
>     Change from v10 to v11
>      * using devm_platform_ioremap_resource
>      * using help macro IMX8_DDR_PMU_EVENT_ATTR
>
>     Change from v9 to v10
>      * use name imx8_ddr_perf_pmu
>      * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
>      * config: 0-7
>      * use read_relaxed() at ddr_perf_read_counter
>      * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
>      * raname ddr_perf_event_start to ddr_perf_counter_enable
>      * use FIELD_PREP
>      * write 0 to disalbe counter at ddr_perf_event_enable
>      * update hwc->state
>      * remove local64_set at ddr_perf_event_add
>      * move perf_pmu_register after irq_register to simplfy err handle
>
>     No change from v8 to v9
>
>     Change from v7 to v8
>      * remove unused define
>      * change total_events to active_events, change active_events to events
>      * remove flags,
>      * fix multi line comments code sytle
>      * add pmu_enable\disable function
>      * disable event at irq handle
>      * remove counter check at ddr_perf_free_counter
>      * remove pmu->irq check
>      * add group check
>
>     Change from v6 to v7
>      * added irq affinity handle, ref arm-ccn.c
>      * added IRQF_NOBALANCING | IRQF_NO_THREAD
>      * added ida_simple_remove at failure path
>
>     Change from v5 to v6
>      * fix insmod\rmmod problem
>      * remove randunt register read at irq handle
>      * change u32 irq to int
>      * devm_request_irq use default flags.
>
>     Change from v4 to v5
>      * Remove AXI ID filter function
>
>     Change from v3 to v4
>      * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
>      * sort include
>      * remove struct fsl_ddr_devtype_data
>      * Added comment need disable control first
>      * Added comment about must enable cycle counter
>      * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
>      * Added comment about cycle counter is fastest one
>
>     Change from v2 to v3
>      * remove kfree
>
>     Change from V1 to V2
>      * update Kconfig by use i.MX8 instead of i.MX8 QXP
>      * remove gpl statememnt since SPDX tag
>      * use dev_kzalloc
>      * use dev_err
>      * commit message show axi_read 0x41\axi_write 0x42
>      * commit message show cycles must be enabled
>      * Irq only issue at cycles overflow
>      * use NUM_COUNTER
>      * use devm_request_irq
>      * add hotplug callback to handle context migration
>
>  drivers/perf/Kconfig             |   7 +
>  drivers/perf/Makefile            |   1 +
>  drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
>  3 files changed, 560 insertions(+)
>  create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c
>
> diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
> index a94e586..9bc3785 100644
> --- a/drivers/perf/Kconfig
> +++ b/drivers/perf/Kconfig
> @@ -70,6 +70,13 @@ config ARM_DSU_PMU
>           system, control logic. The PMU allows counting various events related
>           to DSU.
>
> +config FSL_IMX8_DDR_PMU
> +       tristate "Freescale i.MX8 DDR perf monitor"
> +       depends on ARCH_MXC
> +         help
> +         Provides support for ddr perfomance monitor in i.MX8. Provide memory
> +         througput information.
> +
>  config HISI_PMU
>         bool "HiSilicon SoC PMU"
>         depends on ARM64 && ACPI
> diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
> index 3048994..2ebb4de 100644
> --- a/drivers/perf/Makefile
> +++ b/drivers/perf/Makefile
> @@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
>  obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
>  obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
>  obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
> +obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
>  obj-$(CONFIG_HISI_PMU) += hisilicon/
>  obj-$(CONFIG_QCOM_L2_PMU)      += qcom_l2_pmu.o
>  obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
> diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
> new file mode 100644
> index 0000000..15b0772
> --- /dev/null
> +++ b/drivers/perf/fsl_imx8_ddr_perf.c
> @@ -0,0 +1,552 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright 2017 NXP
> + * Copyright 2016 Freescale Semiconductor, Inc.
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/init.h>
> +#include <linux/interrupt.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/of_irq.h>
> +#include <linux/perf_event.h>
> +#include <linux/slab.h>
> +
> +#define COUNTER_CNTL           0x0
> +#define COUNTER_READ           0x20
> +
> +#define COUNTER_DPCR1          0x30
> +
> +#define CNTL_OVER              0x1
> +#define CNTL_CLEAR             0x2
> +#define CNTL_EN                        0x4
> +#define CNTL_EN_MASK           0xFFFFFFFB
> +#define CNTL_CLEAR_MASK                0xFFFFFFFD
> +#define CNTL_OVER_MASK         0xFFFFFFFE
> +
> +#define CNTL_CSV_SHIFT         24
> +#define CNTL_CSV_MASK          (0xFF << CNTL_CSV_SHIFT)
> +
> +#define EVENT_CYCLES_ID                0
> +#define EVENT_CYCLES_COUNTER   0
> +#define NUM_COUNTERS           4
> +
> +#define to_ddr_pmu(p)          container_of(p, struct ddr_pmu, pmu)
> +
> +#define DDR_PERF_DEV_NAME      "imx8_ddr_perf_pmu"
> +
> +static DEFINE_IDA(ddr_ida);
> +
> +static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
> +       { .compatible = "fsl,imx8-ddr-pmu",},
> +       { .compatible = "fsl,imx8m-ddr-pmu",},
> +       { /* sentinel */ }
> +};
> +
> +struct ddr_pmu {
> +       struct pmu pmu;
> +       void __iomem *base;
> +       unsigned int cpu;
> +       struct  hlist_node node;
> +       struct  device *dev;
> +       struct perf_event *events[NUM_COUNTERS];
> +       int active_events;
> +       enum cpuhp_state cpuhp_state;
> +       int irq;
> +       int id;
> +};
> +
> +static ssize_t ddr_perf_cpumask_show(struct device *dev,
> +                               struct device_attribute *attr, char *buf)
> +{
> +       struct ddr_pmu *pmu = dev_get_drvdata(dev);
> +
> +       return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
> +}
> +
> +static struct device_attribute ddr_perf_cpumask_attr =
> +       __ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
> +
> +static struct attribute *ddr_perf_cpumask_attrs[] = {
> +       &ddr_perf_cpumask_attr.attr,
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_cpumask_attr_group = {
> +       .attrs = ddr_perf_cpumask_attrs,
> +};
> +
> +static ssize_t
> +ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)                            \
> +       (&((struct perf_pmu_events_attr[]) {                            \
> +               { .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
> +                 .id = _id, }                                          \
> +       })[0].attr.attr)
> +
> +static struct attribute *ddr_perf_events_attrs[] = {
> +       IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
> +       IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
> +       IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
> +       IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
> +       IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
> +       IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
> +       IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
> +       IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
> +       IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
> +       IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
> +       IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
> +       IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
> +       IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
> +       IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
> +       IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_events_attr_group = {
> +       .name = "events",
> +       .attrs = ddr_perf_events_attrs,
> +};
> +
> +PMU_FORMAT_ATTR(event, "config:0-7");
> +
> +static struct attribute *ddr_perf_format_attrs[] = {
> +       &format_attr_event.attr,
> +       NULL,
> +};
> +
> +static struct attribute_group ddr_perf_format_attr_group = {
> +       .name = "format",
> +       .attrs = ddr_perf_format_attrs,
> +};
> +
> +static const struct attribute_group *attr_groups[] = {
> +       &ddr_perf_events_attr_group,
> +       &ddr_perf_format_attr_group,
> +       &ddr_perf_cpumask_attr_group,
> +       NULL,
> +};
> +
> +static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
> +{
> +       int i;
> +
> +       /*
> +        * Always map cycle event to counter 0
> +        * Cycles counter is dedicated for cycle event
> +        * can't used for the other events
> +        */
> +       if (event == EVENT_CYCLES_ID) {
> +               if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +                       return EVENT_CYCLES_COUNTER;
> +               else
> +                       return -ENOENT;
> +       }
> +
> +       for (i = 1; i < NUM_COUNTERS; i++) {
> +               if (pmu->events[i] == NULL)
> +                       return i;
> +       }
> +
> +       return -ENOENT;
> +}
> +
> +static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
> +{
> +       pmu->events[counter] = NULL;
> +}
> +
> +static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
> +{
> +       return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
> +}
> +
> +static int ddr_perf_event_init(struct perf_event *event)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       struct perf_event *sibling;
> +
> +       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) {
> +               dev_warn(pmu->dev, "Can't provide per-task data!\n");
> +               return -EOPNOTSUPP;
> +       }
> +
> +       /*
> +        * We must NOT create groups containing mixed PMUs, although software
> +        * events are acceptable (for example to create a CCN group
> +        * periodically read when a hrtimer aka cpu-clock leader triggers).
> +        */
> +       if (event->group_leader->pmu != event->pmu &&
> +                       !is_software_event(event->group_leader))
> +               return -EINVAL;
> +
> +       for_each_sibling_event(sibling, event->group_leader) {
> +               if (sibling->pmu != event->pmu &&
> +                               !is_software_event(sibling))
> +                       return -EINVAL;
> +       }
> +
> +       event->cpu = pmu->cpu;
> +       hwc->idx = -1;
> +
> +       return 0;
> +}
> +
> +
> +static void ddr_perf_event_update(struct perf_event *event)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       u64 delta, prev_raw_count, new_raw_count;
> +       int counter = hwc->idx;
> +
> +       do {
> +               prev_raw_count = local64_read(&hwc->prev_count);
> +               new_raw_count = ddr_perf_read_counter(pmu, counter);
> +       } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
> +                       new_raw_count) != prev_raw_count);
> +
> +       delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
> +
> +       local64_add(delta, &event->count);
> +}
> +
> +static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
> +                                 int counter, bool enable)
> +{
> +       u8 reg = counter * 4 + COUNTER_CNTL;
> +       int val;
> +
> +       if (enable) {
> +               /*
> +                * must disable first, then enable again
> +                * otherwise, cycle counter will not work
> +                * if previous state is enabled.
> +                */
> +               writel(0, pmu->base + reg);
> +               val = CNTL_EN | CNTL_CLEAR;
> +               val |= FIELD_PREP(CNTL_CSV_MASK, config);
> +               writel(val, pmu->base + reg);
> +       } else {
> +               /* Disable counter */
> +               writel(0, pmu->base + reg);
> +       }
> +}
> +
> +static void ddr_perf_event_start(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       local64_set(&hwc->prev_count, 0);
> +
> +       ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
> +
> +       hwc->state = 0;
> +}
> +
> +static int ddr_perf_event_add(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter;
> +       int cfg = event->attr.config;
> +
> +       counter = ddr_perf_alloc_counter(pmu, cfg);
> +       if (counter < 0) {
> +               dev_dbg(pmu->dev, "There are not enough counters\n");
> +               return -EOPNOTSUPP;
> +       }
> +
> +       pmu->events[counter] = event;
> +       pmu->active_events++;
> +       hwc->idx = counter;
> +
> +       hwc->state |= PERF_HES_STOPPED;
> +
> +       if (flags & PERF_EF_START)
> +               ddr_perf_event_start(event, flags);
> +
> +       return 0;
> +}
> +
> +static void ddr_perf_event_stop(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
> +       ddr_perf_event_update(event);
> +
> +       hwc->state |= PERF_HES_STOPPED;
> +}
> +
> +static void ddr_perf_event_del(struct perf_event *event, int flags)
> +{
> +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> +       struct hw_perf_event *hwc = &event->hw;
> +       int counter = hwc->idx;
> +
> +       ddr_perf_event_stop(event, PERF_EF_UPDATE);
> +
> +       ddr_perf_free_counter(pmu, counter);
> +       pmu->active_events--;
> +       hwc->idx = -1;
> +}
> +
> +static void ddr_perf_pmu_enable(struct pmu *pmu)
> +{
> +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> +
> +       /* enable cycle counter if cycle is not active event list */
> +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +               ddr_perf_counter_enable(ddr_pmu,
> +                                     EVENT_CYCLES_ID,
> +                                     EVENT_CYCLES_COUNTER,
> +                                     true);
> +}
> +
> +static void ddr_perf_pmu_disable(struct pmu *pmu)
> +{
> +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> +
> +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> +               ddr_perf_counter_enable(ddr_pmu,
> +                                     EVENT_CYCLES_ID,
> +                                     EVENT_CYCLES_COUNTER,
> +                                     false);
> +}
> +
> +static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
> +                        struct device *dev)
> +{
> +       *pmu = (struct ddr_pmu) {
> +               .pmu = (struct pmu) {
> +                       .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
> +                       .task_ctx_nr = perf_invalid_context,
> +                       .attr_groups = attr_groups,
> +                       .event_init  = ddr_perf_event_init,
> +                       .add         = ddr_perf_event_add,
> +                       .del         = ddr_perf_event_del,
> +                       .start       = ddr_perf_event_start,
> +                       .stop        = ddr_perf_event_stop,
> +                       .read        = ddr_perf_event_update,
> +                       .pmu_enable  = ddr_perf_pmu_enable,
> +                       .pmu_disable = ddr_perf_pmu_disable,
> +               },
> +               .base = base,
> +               .dev = dev,
> +       };
> +
> +       pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
> +       return pmu->id;
> +}
> +
> +static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
> +{
> +       int i;
> +       struct ddr_pmu *pmu = (struct ddr_pmu *) p;
> +       struct perf_event *event, *cycle_event = NULL;
> +
> +       /* all counter will stop if cycle counter disabled */
> +       ddr_perf_counter_enable(pmu,
> +                             EVENT_CYCLES_ID,
> +                             EVENT_CYCLES_COUNTER,
> +                             false);
> +       /*
> +        * When the cycle counter overflows, all counters are stopped,
> +        * and an IRQ is raised. If any other counter overflows, it
> +        * continues counting, and no IRQ is raised.
> +        *
> +        * Cycles occur at least 4 times as often as other events, so we
> +        * can update all events on a cycle counter overflow and not
> +        * lose events.
> +        *
> +        */
> +       for (i = 0; i < NUM_COUNTERS; i++) {
> +
> +               if (!pmu->events[i])
> +                       continue;
> +
> +               event = pmu->events[i];
> +
> +               ddr_perf_event_update(event);
> +
> +               if (event->hw.idx == EVENT_CYCLES_COUNTER)
> +                       cycle_event = event;
> +       }
> +
> +       ddr_perf_counter_enable(pmu,
> +                             EVENT_CYCLES_ID,
> +                             EVENT_CYCLES_COUNTER,
> +                             true);
> +       if (cycle_event)
> +               ddr_perf_event_update(cycle_event);
> +
> +       return IRQ_HANDLED;
> +}
> +
> +static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
> +{
> +       struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
> +       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;
> +
> +       WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
> +
> +       return 0;
> +}
> +
> +static int ddr_perf_probe(struct platform_device *pdev)
> +{
> +       struct ddr_pmu *pmu;
> +       struct device_node *np;
> +       void __iomem *base;
> +       char *name;
> +       int num;
> +       int ret;
> +       int irq;
> +
> +       base = devm_platform_ioremap_resource(pdev, 0);
> +       if (IS_ERR(base))
> +               return PTR_ERR(base);
> +
> +       np = pdev->dev.of_node;
> +
> +       pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
> +       if (!pmu)
> +               return -ENOMEM;
> +
> +       num = ddr_perf_init(pmu, base, &pdev->dev);
> +
> +       platform_set_drvdata(pdev, pmu);
> +
> +       name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
> +       if (!name)
> +               return -ENOMEM;
> +
> +       pmu->cpu = raw_smp_processor_id();
> +       ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> +                                     DDR_PERF_DEV_NAME,
> +                                     NULL,
> +                                     ddr_perf_offline_cpu);
> +
> +       if (ret < 0) {
> +               dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
> +               goto ddr_perf_err;
> +       }
> +
> +       pmu->cpuhp_state = ret;
> +
> +       /* Register the pmu instance for cpu hotplug */
> +       cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +
> +       /* Request irq */
> +       irq = of_irq_get(np, 0);
> +       if (irq < 0) {
> +               dev_err(&pdev->dev, "Failed to get irq: %d", irq);
> +               ret = irq;
> +               goto ddr_perf_err;
> +       }
> +
> +       ret = devm_request_irq(&pdev->dev, irq,
> +                                       ddr_perf_irq_handler,
> +                                       IRQF_NOBALANCING | IRQF_NO_THREAD,
> +                                       DDR_PERF_DEV_NAME,
> +                                       pmu);
> +       if (ret < 0) {
> +               dev_err(&pdev->dev, "Request irq failed: %d", ret);
> +               goto ddr_perf_err;
> +       }
> +
> +       pmu->irq = irq;
> +       ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
> +       if (ret) {
> +               dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
> +               goto ddr_perf_err;
> +       }
> +
> +       ret = perf_pmu_register(&pmu->pmu, name, -1);
> +       if (ret)
> +               goto ddr_perf_err;
> +
> +       return 0;
> +
> +ddr_perf_err:
> +       if (pmu->cpuhp_state)
> +               cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +
> +       ida_simple_remove(&ddr_ida, pmu->id);
> +       dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
> +       return ret;
> +}
> +
> +static int ddr_perf_remove(struct platform_device *pdev)
> +{
> +       struct ddr_pmu *pmu = platform_get_drvdata(pdev);
> +
> +       cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> +       irq_set_affinity_hint(pmu->irq, NULL);
> +
> +       perf_pmu_unregister(&pmu->pmu);
> +
> +       ida_simple_remove(&ddr_ida, pmu->id);
> +       return 0;
> +}
> +
> +static struct platform_driver imx_ddr_pmu_driver = {
> +       .driver         = {
> +               .name   = "imx-ddr-pmu",
> +               .of_match_table = imx_ddr_pmu_dt_ids,
> +       },
> +       .probe          = ddr_perf_probe,
> +       .remove         = ddr_perf_remove,
> +};
> +
> +module_platform_driver(imx_ddr_pmu_driver);
> +MODULE_LICENSE("GPL v2");
> --
> 2.5.2
>

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-05-07 19:30     ` Andrey Smirnov
@ 2019-05-14 13:53       ` Zhi Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-05-14 13:53 UTC (permalink / raw)
  To: Andrey Smirnov
  Cc: mark.rutland, Aisheng Dong, festevam, devicetree, Frank Li,
	s.hauer, will.deacon, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, May 7, 2019 at 2:30 PM Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
>
> On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
> >
> > Add ddr performance monitor support for iMX8QXP
> >
> > There are 4 counters for ddr perfomance events.
> > counter 0 is dedicated for cycles.
> > you choose any up to 3 no cycles events.
> >
> > for example:
> >
> > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> >
> > Support below events.
> >
> >   imx8_ddr0/activate/                                     [Kernel PMU event]
> >   imx8_ddr0/axid-read/                                    [Kernel PMU event]
> >   imx8_ddr0/axid-write/                                   [Kernel PMU event]
> >   imx8_ddr0/cycles/                                       [Kernel PMU event]
> >   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
> >   imx8_ddr0/hp-read/                                      [Kernel PMU event]
> >   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
> >   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/load-mode/                                    [Kernel PMU event]
> >   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
> >   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
> >   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/mwr/                                          [Kernel PMU event]
> >   imx8_ddr0/precharge/                                    [Kernel PMU event]
> >   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
> >   imx8_ddr0/read-access/                                  [Kernel PMU event]
> >   imx8_ddr0/read-activate/                                [Kernel PMU event]
> >   imx8_ddr0/read-command/                                 [Kernel PMU event]
> >   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
> >   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
> >   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
> >   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
> >   imx8_ddr0/read/                                         [Kernel PMU event]
> >   imx8_ddr0/refresh/                                      [Kernel PMU event]
> >   imx8_ddr0/selfresh/                                     [Kernel PMU event]
> >   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/write-access/                                 [Kernel PMU event]
> >   imx8_ddr0/write-command/                                [Kernel PMU event]
> >   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
> >   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
> >   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
> >   imx8_ddr0/write/
> >
> > Signed-off-by: Frank Li <Frank.Li@nxp.com>
>
> Don't really have anything to suggest, that I already haven't. LGTM, so:
>
> Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>


Will Deacon:
            Is it okay to pick up these patches?  Andrey always acked.

best regards
Frank Li

>
>
> > ---
> >
> > Notes:
> >     Change from v11 to v12
> >     * change name from imx_ddr to imx8_ddr
> >
> >     Change from v10 to v11
> >      * using devm_platform_ioremap_resource
> >      * using help macro IMX8_DDR_PMU_EVENT_ATTR
> >
> >     Change from v9 to v10
> >      * use name imx8_ddr_perf_pmu
> >      * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
> >      * config: 0-7
> >      * use read_relaxed() at ddr_perf_read_counter
> >      * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
> >      * raname ddr_perf_event_start to ddr_perf_counter_enable
> >      * use FIELD_PREP
> >      * write 0 to disalbe counter at ddr_perf_event_enable
> >      * update hwc->state
> >      * remove local64_set at ddr_perf_event_add
> >      * move perf_pmu_register after irq_register to simplfy err handle
> >
> >     No change from v8 to v9
> >
> >     Change from v7 to v8
> >      * remove unused define
> >      * change total_events to active_events, change active_events to events
> >      * remove flags,
> >      * fix multi line comments code sytle
> >      * add pmu_enable\disable function
> >      * disable event at irq handle
> >      * remove counter check at ddr_perf_free_counter
> >      * remove pmu->irq check
> >      * add group check
> >
> >     Change from v6 to v7
> >      * added irq affinity handle, ref arm-ccn.c
> >      * added IRQF_NOBALANCING | IRQF_NO_THREAD
> >      * added ida_simple_remove at failure path
> >
> >     Change from v5 to v6
> >      * fix insmod\rmmod problem
> >      * remove randunt register read at irq handle
> >      * change u32 irq to int
> >      * devm_request_irq use default flags.
> >
> >     Change from v4 to v5
> >      * Remove AXI ID filter function
> >
> >     Change from v3 to v4
> >      * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
> >      * sort include
> >      * remove struct fsl_ddr_devtype_data
> >      * Added comment need disable control first
> >      * Added comment about must enable cycle counter
> >      * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
> >      * Added comment about cycle counter is fastest one
> >
> >     Change from v2 to v3
> >      * remove kfree
> >
> >     Change from V1 to V2
> >      * update Kconfig by use i.MX8 instead of i.MX8 QXP
> >      * remove gpl statememnt since SPDX tag
> >      * use dev_kzalloc
> >      * use dev_err
> >      * commit message show axi_read 0x41\axi_write 0x42
> >      * commit message show cycles must be enabled
> >      * Irq only issue at cycles overflow
> >      * use NUM_COUNTER
> >      * use devm_request_irq
> >      * add hotplug callback to handle context migration
> >
> >  drivers/perf/Kconfig             |   7 +
> >  drivers/perf/Makefile            |   1 +
> >  drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 560 insertions(+)
> >  create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c
> >
> > diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
> > index a94e586..9bc3785 100644
> > --- a/drivers/perf/Kconfig
> > +++ b/drivers/perf/Kconfig
> > @@ -70,6 +70,13 @@ config ARM_DSU_PMU
> >           system, control logic. The PMU allows counting various events related
> >           to DSU.
> >
> > +config FSL_IMX8_DDR_PMU
> > +       tristate "Freescale i.MX8 DDR perf monitor"
> > +       depends on ARCH_MXC
> > +         help
> > +         Provides support for ddr perfomance monitor in i.MX8. Provide memory
> > +         througput information.
> > +
> >  config HISI_PMU
> >         bool "HiSilicon SoC PMU"
> >         depends on ARM64 && ACPI
> > diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
> > index 3048994..2ebb4de 100644
> > --- a/drivers/perf/Makefile
> > +++ b/drivers/perf/Makefile
> > @@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
> >  obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
> >  obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
> >  obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
> > +obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
> >  obj-$(CONFIG_HISI_PMU) += hisilicon/
> >  obj-$(CONFIG_QCOM_L2_PMU)      += qcom_l2_pmu.o
> >  obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
> > diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
> > new file mode 100644
> > index 0000000..15b0772
> > --- /dev/null
> > +++ b/drivers/perf/fsl_imx8_ddr_perf.c
> > @@ -0,0 +1,552 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright 2017 NXP
> > + * Copyright 2016 Freescale Semiconductor, Inc.
> > + */
> > +
> > +#include <linux/bitfield.h>
> > +#include <linux/init.h>
> > +#include <linux/interrupt.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/of_irq.h>
> > +#include <linux/perf_event.h>
> > +#include <linux/slab.h>
> > +
> > +#define COUNTER_CNTL           0x0
> > +#define COUNTER_READ           0x20
> > +
> > +#define COUNTER_DPCR1          0x30
> > +
> > +#define CNTL_OVER              0x1
> > +#define CNTL_CLEAR             0x2
> > +#define CNTL_EN                        0x4
> > +#define CNTL_EN_MASK           0xFFFFFFFB
> > +#define CNTL_CLEAR_MASK                0xFFFFFFFD
> > +#define CNTL_OVER_MASK         0xFFFFFFFE
> > +
> > +#define CNTL_CSV_SHIFT         24
> > +#define CNTL_CSV_MASK          (0xFF << CNTL_CSV_SHIFT)
> > +
> > +#define EVENT_CYCLES_ID                0
> > +#define EVENT_CYCLES_COUNTER   0
> > +#define NUM_COUNTERS           4
> > +
> > +#define to_ddr_pmu(p)          container_of(p, struct ddr_pmu, pmu)
> > +
> > +#define DDR_PERF_DEV_NAME      "imx8_ddr_perf_pmu"
> > +
> > +static DEFINE_IDA(ddr_ida);
> > +
> > +static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
> > +       { .compatible = "fsl,imx8-ddr-pmu",},
> > +       { .compatible = "fsl,imx8m-ddr-pmu",},
> > +       { /* sentinel */ }
> > +};
> > +
> > +struct ddr_pmu {
> > +       struct pmu pmu;
> > +       void __iomem *base;
> > +       unsigned int cpu;
> > +       struct  hlist_node node;
> > +       struct  device *dev;
> > +       struct perf_event *events[NUM_COUNTERS];
> > +       int active_events;
> > +       enum cpuhp_state cpuhp_state;
> > +       int irq;
> > +       int id;
> > +};
> > +
> > +static ssize_t ddr_perf_cpumask_show(struct device *dev,
> > +                               struct device_attribute *attr, char *buf)
> > +{
> > +       struct ddr_pmu *pmu = dev_get_drvdata(dev);
> > +
> > +       return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
> > +}
> > +
> > +static struct device_attribute ddr_perf_cpumask_attr =
> > +       __ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
> > +
> > +static struct attribute *ddr_perf_cpumask_attrs[] = {
> > +       &ddr_perf_cpumask_attr.attr,
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_cpumask_attr_group = {
> > +       .attrs = ddr_perf_cpumask_attrs,
> > +};
> > +
> > +static ssize_t
> > +ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)                            \
> > +       (&((struct perf_pmu_events_attr[]) {                            \
> > +               { .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
> > +                 .id = _id, }                                          \
> > +       })[0].attr.attr)
> > +
> > +static struct attribute *ddr_perf_events_attrs[] = {
> > +       IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
> > +       IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
> > +       IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
> > +       IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
> > +       IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
> > +       IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
> > +       IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
> > +       IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
> > +       IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_events_attr_group = {
> > +       .name = "events",
> > +       .attrs = ddr_perf_events_attrs,
> > +};
> > +
> > +PMU_FORMAT_ATTR(event, "config:0-7");
> > +
> > +static struct attribute *ddr_perf_format_attrs[] = {
> > +       &format_attr_event.attr,
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_format_attr_group = {
> > +       .name = "format",
> > +       .attrs = ddr_perf_format_attrs,
> > +};
> > +
> > +static const struct attribute_group *attr_groups[] = {
> > +       &ddr_perf_events_attr_group,
> > +       &ddr_perf_format_attr_group,
> > +       &ddr_perf_cpumask_attr_group,
> > +       NULL,
> > +};
> > +
> > +static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
> > +{
> > +       int i;
> > +
> > +       /*
> > +        * Always map cycle event to counter 0
> > +        * Cycles counter is dedicated for cycle event
> > +        * can't used for the other events
> > +        */
> > +       if (event == EVENT_CYCLES_ID) {
> > +               if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +                       return EVENT_CYCLES_COUNTER;
> > +               else
> > +                       return -ENOENT;
> > +       }
> > +
> > +       for (i = 1; i < NUM_COUNTERS; i++) {
> > +               if (pmu->events[i] == NULL)
> > +                       return i;
> > +       }
> > +
> > +       return -ENOENT;
> > +}
> > +
> > +static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
> > +{
> > +       pmu->events[counter] = NULL;
> > +}
> > +
> > +static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
> > +{
> > +       return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
> > +}
> > +
> > +static int ddr_perf_event_init(struct perf_event *event)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       struct perf_event *sibling;
> > +
> > +       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) {
> > +               dev_warn(pmu->dev, "Can't provide per-task data!\n");
> > +               return -EOPNOTSUPP;
> > +       }
> > +
> > +       /*
> > +        * We must NOT create groups containing mixed PMUs, although software
> > +        * events are acceptable (for example to create a CCN group
> > +        * periodically read when a hrtimer aka cpu-clock leader triggers).
> > +        */
> > +       if (event->group_leader->pmu != event->pmu &&
> > +                       !is_software_event(event->group_leader))
> > +               return -EINVAL;
> > +
> > +       for_each_sibling_event(sibling, event->group_leader) {
> > +               if (sibling->pmu != event->pmu &&
> > +                               !is_software_event(sibling))
> > +                       return -EINVAL;
> > +       }
> > +
> > +       event->cpu = pmu->cpu;
> > +       hwc->idx = -1;
> > +
> > +       return 0;
> > +}
> > +
> > +
> > +static void ddr_perf_event_update(struct perf_event *event)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       u64 delta, prev_raw_count, new_raw_count;
> > +       int counter = hwc->idx;
> > +
> > +       do {
> > +               prev_raw_count = local64_read(&hwc->prev_count);
> > +               new_raw_count = ddr_perf_read_counter(pmu, counter);
> > +       } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
> > +                       new_raw_count) != prev_raw_count);
> > +
> > +       delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
> > +
> > +       local64_add(delta, &event->count);
> > +}
> > +
> > +static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
> > +                                 int counter, bool enable)
> > +{
> > +       u8 reg = counter * 4 + COUNTER_CNTL;
> > +       int val;
> > +
> > +       if (enable) {
> > +               /*
> > +                * must disable first, then enable again
> > +                * otherwise, cycle counter will not work
> > +                * if previous state is enabled.
> > +                */
> > +               writel(0, pmu->base + reg);
> > +               val = CNTL_EN | CNTL_CLEAR;
> > +               val |= FIELD_PREP(CNTL_CSV_MASK, config);
> > +               writel(val, pmu->base + reg);
> > +       } else {
> > +               /* Disable counter */
> > +               writel(0, pmu->base + reg);
> > +       }
> > +}
> > +
> > +static void ddr_perf_event_start(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       local64_set(&hwc->prev_count, 0);
> > +
> > +       ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
> > +
> > +       hwc->state = 0;
> > +}
> > +
> > +static int ddr_perf_event_add(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter;
> > +       int cfg = event->attr.config;
> > +
> > +       counter = ddr_perf_alloc_counter(pmu, cfg);
> > +       if (counter < 0) {
> > +               dev_dbg(pmu->dev, "There are not enough counters\n");
> > +               return -EOPNOTSUPP;
> > +       }
> > +
> > +       pmu->events[counter] = event;
> > +       pmu->active_events++;
> > +       hwc->idx = counter;
> > +
> > +       hwc->state |= PERF_HES_STOPPED;
> > +
> > +       if (flags & PERF_EF_START)
> > +               ddr_perf_event_start(event, flags);
> > +
> > +       return 0;
> > +}
> > +
> > +static void ddr_perf_event_stop(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
> > +       ddr_perf_event_update(event);
> > +
> > +       hwc->state |= PERF_HES_STOPPED;
> > +}
> > +
> > +static void ddr_perf_event_del(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       ddr_perf_event_stop(event, PERF_EF_UPDATE);
> > +
> > +       ddr_perf_free_counter(pmu, counter);
> > +       pmu->active_events--;
> > +       hwc->idx = -1;
> > +}
> > +
> > +static void ddr_perf_pmu_enable(struct pmu *pmu)
> > +{
> > +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> > +
> > +       /* enable cycle counter if cycle is not active event list */
> > +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +               ddr_perf_counter_enable(ddr_pmu,
> > +                                     EVENT_CYCLES_ID,
> > +                                     EVENT_CYCLES_COUNTER,
> > +                                     true);
> > +}
> > +
> > +static void ddr_perf_pmu_disable(struct pmu *pmu)
> > +{
> > +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> > +
> > +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +               ddr_perf_counter_enable(ddr_pmu,
> > +                                     EVENT_CYCLES_ID,
> > +                                     EVENT_CYCLES_COUNTER,
> > +                                     false);
> > +}
> > +
> > +static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
> > +                        struct device *dev)
> > +{
> > +       *pmu = (struct ddr_pmu) {
> > +               .pmu = (struct pmu) {
> > +                       .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
> > +                       .task_ctx_nr = perf_invalid_context,
> > +                       .attr_groups = attr_groups,
> > +                       .event_init  = ddr_perf_event_init,
> > +                       .add         = ddr_perf_event_add,
> > +                       .del         = ddr_perf_event_del,
> > +                       .start       = ddr_perf_event_start,
> > +                       .stop        = ddr_perf_event_stop,
> > +                       .read        = ddr_perf_event_update,
> > +                       .pmu_enable  = ddr_perf_pmu_enable,
> > +                       .pmu_disable = ddr_perf_pmu_disable,
> > +               },
> > +               .base = base,
> > +               .dev = dev,
> > +       };
> > +
> > +       pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
> > +       return pmu->id;
> > +}
> > +
> > +static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
> > +{
> > +       int i;
> > +       struct ddr_pmu *pmu = (struct ddr_pmu *) p;
> > +       struct perf_event *event, *cycle_event = NULL;
> > +
> > +       /* all counter will stop if cycle counter disabled */
> > +       ddr_perf_counter_enable(pmu,
> > +                             EVENT_CYCLES_ID,
> > +                             EVENT_CYCLES_COUNTER,
> > +                             false);
> > +       /*
> > +        * When the cycle counter overflows, all counters are stopped,
> > +        * and an IRQ is raised. If any other counter overflows, it
> > +        * continues counting, and no IRQ is raised.
> > +        *
> > +        * Cycles occur at least 4 times as often as other events, so we
> > +        * can update all events on a cycle counter overflow and not
> > +        * lose events.
> > +        *
> > +        */
> > +       for (i = 0; i < NUM_COUNTERS; i++) {
> > +
> > +               if (!pmu->events[i])
> > +                       continue;
> > +
> > +               event = pmu->events[i];
> > +
> > +               ddr_perf_event_update(event);
> > +
> > +               if (event->hw.idx == EVENT_CYCLES_COUNTER)
> > +                       cycle_event = event;
> > +       }
> > +
> > +       ddr_perf_counter_enable(pmu,
> > +                             EVENT_CYCLES_ID,
> > +                             EVENT_CYCLES_COUNTER,
> > +                             true);
> > +       if (cycle_event)
> > +               ddr_perf_event_update(cycle_event);
> > +
> > +       return IRQ_HANDLED;
> > +}
> > +
> > +static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
> > +{
> > +       struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
> > +       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;
> > +
> > +       WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
> > +
> > +       return 0;
> > +}
> > +
> > +static int ddr_perf_probe(struct platform_device *pdev)
> > +{
> > +       struct ddr_pmu *pmu;
> > +       struct device_node *np;
> > +       void __iomem *base;
> > +       char *name;
> > +       int num;
> > +       int ret;
> > +       int irq;
> > +
> > +       base = devm_platform_ioremap_resource(pdev, 0);
> > +       if (IS_ERR(base))
> > +               return PTR_ERR(base);
> > +
> > +       np = pdev->dev.of_node;
> > +
> > +       pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
> > +       if (!pmu)
> > +               return -ENOMEM;
> > +
> > +       num = ddr_perf_init(pmu, base, &pdev->dev);
> > +
> > +       platform_set_drvdata(pdev, pmu);
> > +
> > +       name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
> > +       if (!name)
> > +               return -ENOMEM;
> > +
> > +       pmu->cpu = raw_smp_processor_id();
> > +       ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> > +                                     DDR_PERF_DEV_NAME,
> > +                                     NULL,
> > +                                     ddr_perf_offline_cpu);
> > +
> > +       if (ret < 0) {
> > +               dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       pmu->cpuhp_state = ret;
> > +
> > +       /* Register the pmu instance for cpu hotplug */
> > +       cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +
> > +       /* Request irq */
> > +       irq = of_irq_get(np, 0);
> > +       if (irq < 0) {
> > +               dev_err(&pdev->dev, "Failed to get irq: %d", irq);
> > +               ret = irq;
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       ret = devm_request_irq(&pdev->dev, irq,
> > +                                       ddr_perf_irq_handler,
> > +                                       IRQF_NOBALANCING | IRQF_NO_THREAD,
> > +                                       DDR_PERF_DEV_NAME,
> > +                                       pmu);
> > +       if (ret < 0) {
> > +               dev_err(&pdev->dev, "Request irq failed: %d", ret);
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       pmu->irq = irq;
> > +       ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
> > +       if (ret) {
> > +               dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       ret = perf_pmu_register(&pmu->pmu, name, -1);
> > +       if (ret)
> > +               goto ddr_perf_err;
> > +
> > +       return 0;
> > +
> > +ddr_perf_err:
> > +       if (pmu->cpuhp_state)
> > +               cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +
> > +       ida_simple_remove(&ddr_ida, pmu->id);
> > +       dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
> > +       return ret;
> > +}
> > +
> > +static int ddr_perf_remove(struct platform_device *pdev)
> > +{
> > +       struct ddr_pmu *pmu = platform_get_drvdata(pdev);
> > +
> > +       cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +       irq_set_affinity_hint(pmu->irq, NULL);
> > +
> > +       perf_pmu_unregister(&pmu->pmu);
> > +
> > +       ida_simple_remove(&ddr_ida, pmu->id);
> > +       return 0;
> > +}
> > +
> > +static struct platform_driver imx_ddr_pmu_driver = {
> > +       .driver         = {
> > +               .name   = "imx-ddr-pmu",
> > +               .of_match_table = imx_ddr_pmu_dt_ids,
> > +       },
> > +       .probe          = ddr_perf_probe,
> > +       .remove         = ddr_perf_remove,
> > +};
> > +
> > +module_platform_driver(imx_ddr_pmu_driver);
> > +MODULE_LICENSE("GPL v2");
> > --
> > 2.5.2
> >

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-05-14 13:53       ` Zhi Li
  0 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-05-14 13:53 UTC (permalink / raw)
  To: Andrey Smirnov
  Cc: mark.rutland, Aisheng Dong, festevam, devicetree, Frank Li,
	s.hauer, will.deacon, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, May 7, 2019 at 2:30 PM Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
>
> On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
> >
> > Add ddr performance monitor support for iMX8QXP
> >
> > There are 4 counters for ddr perfomance events.
> > counter 0 is dedicated for cycles.
> > you choose any up to 3 no cycles events.
> >
> > for example:
> >
> > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> >
> > Support below events.
> >
> >   imx8_ddr0/activate/                                     [Kernel PMU event]
> >   imx8_ddr0/axid-read/                                    [Kernel PMU event]
> >   imx8_ddr0/axid-write/                                   [Kernel PMU event]
> >   imx8_ddr0/cycles/                                       [Kernel PMU event]
> >   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
> >   imx8_ddr0/hp-read/                                      [Kernel PMU event]
> >   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
> >   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/load-mode/                                    [Kernel PMU event]
> >   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
> >   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
> >   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/mwr/                                          [Kernel PMU event]
> >   imx8_ddr0/precharge/                                    [Kernel PMU event]
> >   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
> >   imx8_ddr0/read-access/                                  [Kernel PMU event]
> >   imx8_ddr0/read-activate/                                [Kernel PMU event]
> >   imx8_ddr0/read-command/                                 [Kernel PMU event]
> >   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
> >   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
> >   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
> >   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
> >   imx8_ddr0/read/                                         [Kernel PMU event]
> >   imx8_ddr0/refresh/                                      [Kernel PMU event]
> >   imx8_ddr0/selfresh/                                     [Kernel PMU event]
> >   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
> >   imx8_ddr0/write-access/                                 [Kernel PMU event]
> >   imx8_ddr0/write-command/                                [Kernel PMU event]
> >   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
> >   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
> >   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
> >   imx8_ddr0/write/
> >
> > Signed-off-by: Frank Li <Frank.Li@nxp.com>
>
> Don't really have anything to suggest, that I already haven't. LGTM, so:
>
> Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>


Will Deacon:
            Is it okay to pick up these patches?  Andrey always acked.

best regards
Frank Li

>
>
> > ---
> >
> > Notes:
> >     Change from v11 to v12
> >     * change name from imx_ddr to imx8_ddr
> >
> >     Change from v10 to v11
> >      * using devm_platform_ioremap_resource
> >      * using help macro IMX8_DDR_PMU_EVENT_ATTR
> >
> >     Change from v9 to v10
> >      * use name imx8_ddr_perf_pmu
> >      * cpuhp_setup_state_multi use DDR_PERF_DEV_NAME
> >      * config: 0-7
> >      * use read_relaxed() at ddr_perf_read_counter
> >      * add .capabilities = PERF_PMU_CAP_NO_EXCLUDE
> >      * raname ddr_perf_event_start to ddr_perf_counter_enable
> >      * use FIELD_PREP
> >      * write 0 to disalbe counter at ddr_perf_event_enable
> >      * update hwc->state
> >      * remove local64_set at ddr_perf_event_add
> >      * move perf_pmu_register after irq_register to simplfy err handle
> >
> >     No change from v8 to v9
> >
> >     Change from v7 to v8
> >      * remove unused define
> >      * change total_events to active_events, change active_events to events
> >      * remove flags,
> >      * fix multi line comments code sytle
> >      * add pmu_enable\disable function
> >      * disable event at irq handle
> >      * remove counter check at ddr_perf_free_counter
> >      * remove pmu->irq check
> >      * add group check
> >
> >     Change from v6 to v7
> >      * added irq affinity handle, ref arm-ccn.c
> >      * added IRQF_NOBALANCING | IRQF_NO_THREAD
> >      * added ida_simple_remove at failure path
> >
> >     Change from v5 to v6
> >      * fix insmod\rmmod problem
> >      * remove randunt register read at irq handle
> >      * change u32 irq to int
> >      * devm_request_irq use default flags.
> >
> >     Change from v4 to v5
> >      * Remove AXI ID filter function
> >
> >     Change from v3 to v4
> >      * Change FSL_IMX8_DDR_PERF to FSL_IMX8_DDR_PMU
> >      * sort include
> >      * remove struct fsl_ddr_devtype_data
> >      * Added comment need disable control first
> >      * Added comment about must enable cycle counter
> >      * Added macro for EVENT_AXI_READ, remove hardcode 0x41 and 0x42
> >      * Added comment about cycle counter is fastest one
> >
> >     Change from v2 to v3
> >      * remove kfree
> >
> >     Change from V1 to V2
> >      * update Kconfig by use i.MX8 instead of i.MX8 QXP
> >      * remove gpl statememnt since SPDX tag
> >      * use dev_kzalloc
> >      * use dev_err
> >      * commit message show axi_read 0x41\axi_write 0x42
> >      * commit message show cycles must be enabled
> >      * Irq only issue at cycles overflow
> >      * use NUM_COUNTER
> >      * use devm_request_irq
> >      * add hotplug callback to handle context migration
> >
> >  drivers/perf/Kconfig             |   7 +
> >  drivers/perf/Makefile            |   1 +
> >  drivers/perf/fsl_imx8_ddr_perf.c | 552 +++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 560 insertions(+)
> >  create mode 100644 drivers/perf/fsl_imx8_ddr_perf.c
> >
> > diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
> > index a94e586..9bc3785 100644
> > --- a/drivers/perf/Kconfig
> > +++ b/drivers/perf/Kconfig
> > @@ -70,6 +70,13 @@ config ARM_DSU_PMU
> >           system, control logic. The PMU allows counting various events related
> >           to DSU.
> >
> > +config FSL_IMX8_DDR_PMU
> > +       tristate "Freescale i.MX8 DDR perf monitor"
> > +       depends on ARCH_MXC
> > +         help
> > +         Provides support for ddr perfomance monitor in i.MX8. Provide memory
> > +         througput information.
> > +
> >  config HISI_PMU
> >         bool "HiSilicon SoC PMU"
> >         depends on ARM64 && ACPI
> > diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
> > index 3048994..2ebb4de 100644
> > --- a/drivers/perf/Makefile
> > +++ b/drivers/perf/Makefile
> > @@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_DSU_PMU) += arm_dsu_pmu.o
> >  obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
> >  obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
> >  obj-$(CONFIG_ARM_SMMU_V3_PMU) += arm_smmuv3_pmu.o
> > +obj-$(CONFIG_FSL_IMX8_DDR_PMU) += fsl_imx8_ddr_perf.o
> >  obj-$(CONFIG_HISI_PMU) += hisilicon/
> >  obj-$(CONFIG_QCOM_L2_PMU)      += qcom_l2_pmu.o
> >  obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
> > diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c
> > new file mode 100644
> > index 0000000..15b0772
> > --- /dev/null
> > +++ b/drivers/perf/fsl_imx8_ddr_perf.c
> > @@ -0,0 +1,552 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright 2017 NXP
> > + * Copyright 2016 Freescale Semiconductor, Inc.
> > + */
> > +
> > +#include <linux/bitfield.h>
> > +#include <linux/init.h>
> > +#include <linux/interrupt.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/of_irq.h>
> > +#include <linux/perf_event.h>
> > +#include <linux/slab.h>
> > +
> > +#define COUNTER_CNTL           0x0
> > +#define COUNTER_READ           0x20
> > +
> > +#define COUNTER_DPCR1          0x30
> > +
> > +#define CNTL_OVER              0x1
> > +#define CNTL_CLEAR             0x2
> > +#define CNTL_EN                        0x4
> > +#define CNTL_EN_MASK           0xFFFFFFFB
> > +#define CNTL_CLEAR_MASK                0xFFFFFFFD
> > +#define CNTL_OVER_MASK         0xFFFFFFFE
> > +
> > +#define CNTL_CSV_SHIFT         24
> > +#define CNTL_CSV_MASK          (0xFF << CNTL_CSV_SHIFT)
> > +
> > +#define EVENT_CYCLES_ID                0
> > +#define EVENT_CYCLES_COUNTER   0
> > +#define NUM_COUNTERS           4
> > +
> > +#define to_ddr_pmu(p)          container_of(p, struct ddr_pmu, pmu)
> > +
> > +#define DDR_PERF_DEV_NAME      "imx8_ddr_perf_pmu"
> > +
> > +static DEFINE_IDA(ddr_ida);
> > +
> > +static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
> > +       { .compatible = "fsl,imx8-ddr-pmu",},
> > +       { .compatible = "fsl,imx8m-ddr-pmu",},
> > +       { /* sentinel */ }
> > +};
> > +
> > +struct ddr_pmu {
> > +       struct pmu pmu;
> > +       void __iomem *base;
> > +       unsigned int cpu;
> > +       struct  hlist_node node;
> > +       struct  device *dev;
> > +       struct perf_event *events[NUM_COUNTERS];
> > +       int active_events;
> > +       enum cpuhp_state cpuhp_state;
> > +       int irq;
> > +       int id;
> > +};
> > +
> > +static ssize_t ddr_perf_cpumask_show(struct device *dev,
> > +                               struct device_attribute *attr, char *buf)
> > +{
> > +       struct ddr_pmu *pmu = dev_get_drvdata(dev);
> > +
> > +       return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
> > +}
> > +
> > +static struct device_attribute ddr_perf_cpumask_attr =
> > +       __ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);
> > +
> > +static struct attribute *ddr_perf_cpumask_attrs[] = {
> > +       &ddr_perf_cpumask_attr.attr,
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_cpumask_attr_group = {
> > +       .attrs = ddr_perf_cpumask_attrs,
> > +};
> > +
> > +static ssize_t
> > +ddr_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 IMX8_DDR_PMU_EVENT_ATTR(_name, _id)                            \
> > +       (&((struct perf_pmu_events_attr[]) {                            \
> > +               { .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
> > +                 .id = _id, }                                          \
> > +       })[0].attr.attr)
> > +
> > +static struct attribute *ddr_perf_events_attrs[] = {
> > +       IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
> > +       IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
> > +       IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
> > +       IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
> > +       IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
> > +       IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
> > +       IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
> > +       IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
> > +       IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
> > +       IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
> > +       IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
> > +       IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
> > +       IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_events_attr_group = {
> > +       .name = "events",
> > +       .attrs = ddr_perf_events_attrs,
> > +};
> > +
> > +PMU_FORMAT_ATTR(event, "config:0-7");
> > +
> > +static struct attribute *ddr_perf_format_attrs[] = {
> > +       &format_attr_event.attr,
> > +       NULL,
> > +};
> > +
> > +static struct attribute_group ddr_perf_format_attr_group = {
> > +       .name = "format",
> > +       .attrs = ddr_perf_format_attrs,
> > +};
> > +
> > +static const struct attribute_group *attr_groups[] = {
> > +       &ddr_perf_events_attr_group,
> > +       &ddr_perf_format_attr_group,
> > +       &ddr_perf_cpumask_attr_group,
> > +       NULL,
> > +};
> > +
> > +static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
> > +{
> > +       int i;
> > +
> > +       /*
> > +        * Always map cycle event to counter 0
> > +        * Cycles counter is dedicated for cycle event
> > +        * can't used for the other events
> > +        */
> > +       if (event == EVENT_CYCLES_ID) {
> > +               if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +                       return EVENT_CYCLES_COUNTER;
> > +               else
> > +                       return -ENOENT;
> > +       }
> > +
> > +       for (i = 1; i < NUM_COUNTERS; i++) {
> > +               if (pmu->events[i] == NULL)
> > +                       return i;
> > +       }
> > +
> > +       return -ENOENT;
> > +}
> > +
> > +static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
> > +{
> > +       pmu->events[counter] = NULL;
> > +}
> > +
> > +static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
> > +{
> > +       return readl_relaxed(pmu->base + COUNTER_READ + counter * 4);
> > +}
> > +
> > +static int ddr_perf_event_init(struct perf_event *event)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       struct perf_event *sibling;
> > +
> > +       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) {
> > +               dev_warn(pmu->dev, "Can't provide per-task data!\n");
> > +               return -EOPNOTSUPP;
> > +       }
> > +
> > +       /*
> > +        * We must NOT create groups containing mixed PMUs, although software
> > +        * events are acceptable (for example to create a CCN group
> > +        * periodically read when a hrtimer aka cpu-clock leader triggers).
> > +        */
> > +       if (event->group_leader->pmu != event->pmu &&
> > +                       !is_software_event(event->group_leader))
> > +               return -EINVAL;
> > +
> > +       for_each_sibling_event(sibling, event->group_leader) {
> > +               if (sibling->pmu != event->pmu &&
> > +                               !is_software_event(sibling))
> > +                       return -EINVAL;
> > +       }
> > +
> > +       event->cpu = pmu->cpu;
> > +       hwc->idx = -1;
> > +
> > +       return 0;
> > +}
> > +
> > +
> > +static void ddr_perf_event_update(struct perf_event *event)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       u64 delta, prev_raw_count, new_raw_count;
> > +       int counter = hwc->idx;
> > +
> > +       do {
> > +               prev_raw_count = local64_read(&hwc->prev_count);
> > +               new_raw_count = ddr_perf_read_counter(pmu, counter);
> > +       } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
> > +                       new_raw_count) != prev_raw_count);
> > +
> > +       delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
> > +
> > +       local64_add(delta, &event->count);
> > +}
> > +
> > +static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
> > +                                 int counter, bool enable)
> > +{
> > +       u8 reg = counter * 4 + COUNTER_CNTL;
> > +       int val;
> > +
> > +       if (enable) {
> > +               /*
> > +                * must disable first, then enable again
> > +                * otherwise, cycle counter will not work
> > +                * if previous state is enabled.
> > +                */
> > +               writel(0, pmu->base + reg);
> > +               val = CNTL_EN | CNTL_CLEAR;
> > +               val |= FIELD_PREP(CNTL_CSV_MASK, config);
> > +               writel(val, pmu->base + reg);
> > +       } else {
> > +               /* Disable counter */
> > +               writel(0, pmu->base + reg);
> > +       }
> > +}
> > +
> > +static void ddr_perf_event_start(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       local64_set(&hwc->prev_count, 0);
> > +
> > +       ddr_perf_counter_enable(pmu, event->attr.config, counter, true);
> > +
> > +       hwc->state = 0;
> > +}
> > +
> > +static int ddr_perf_event_add(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter;
> > +       int cfg = event->attr.config;
> > +
> > +       counter = ddr_perf_alloc_counter(pmu, cfg);
> > +       if (counter < 0) {
> > +               dev_dbg(pmu->dev, "There are not enough counters\n");
> > +               return -EOPNOTSUPP;
> > +       }
> > +
> > +       pmu->events[counter] = event;
> > +       pmu->active_events++;
> > +       hwc->idx = counter;
> > +
> > +       hwc->state |= PERF_HES_STOPPED;
> > +
> > +       if (flags & PERF_EF_START)
> > +               ddr_perf_event_start(event, flags);
> > +
> > +       return 0;
> > +}
> > +
> > +static void ddr_perf_event_stop(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
> > +       ddr_perf_event_update(event);
> > +
> > +       hwc->state |= PERF_HES_STOPPED;
> > +}
> > +
> > +static void ddr_perf_event_del(struct perf_event *event, int flags)
> > +{
> > +       struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
> > +       struct hw_perf_event *hwc = &event->hw;
> > +       int counter = hwc->idx;
> > +
> > +       ddr_perf_event_stop(event, PERF_EF_UPDATE);
> > +
> > +       ddr_perf_free_counter(pmu, counter);
> > +       pmu->active_events--;
> > +       hwc->idx = -1;
> > +}
> > +
> > +static void ddr_perf_pmu_enable(struct pmu *pmu)
> > +{
> > +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> > +
> > +       /* enable cycle counter if cycle is not active event list */
> > +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +               ddr_perf_counter_enable(ddr_pmu,
> > +                                     EVENT_CYCLES_ID,
> > +                                     EVENT_CYCLES_COUNTER,
> > +                                     true);
> > +}
> > +
> > +static void ddr_perf_pmu_disable(struct pmu *pmu)
> > +{
> > +       struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);
> > +
> > +       if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
> > +               ddr_perf_counter_enable(ddr_pmu,
> > +                                     EVENT_CYCLES_ID,
> > +                                     EVENT_CYCLES_COUNTER,
> > +                                     false);
> > +}
> > +
> > +static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
> > +                        struct device *dev)
> > +{
> > +       *pmu = (struct ddr_pmu) {
> > +               .pmu = (struct pmu) {
> > +                       .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
> > +                       .task_ctx_nr = perf_invalid_context,
> > +                       .attr_groups = attr_groups,
> > +                       .event_init  = ddr_perf_event_init,
> > +                       .add         = ddr_perf_event_add,
> > +                       .del         = ddr_perf_event_del,
> > +                       .start       = ddr_perf_event_start,
> > +                       .stop        = ddr_perf_event_stop,
> > +                       .read        = ddr_perf_event_update,
> > +                       .pmu_enable  = ddr_perf_pmu_enable,
> > +                       .pmu_disable = ddr_perf_pmu_disable,
> > +               },
> > +               .base = base,
> > +               .dev = dev,
> > +       };
> > +
> > +       pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
> > +       return pmu->id;
> > +}
> > +
> > +static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
> > +{
> > +       int i;
> > +       struct ddr_pmu *pmu = (struct ddr_pmu *) p;
> > +       struct perf_event *event, *cycle_event = NULL;
> > +
> > +       /* all counter will stop if cycle counter disabled */
> > +       ddr_perf_counter_enable(pmu,
> > +                             EVENT_CYCLES_ID,
> > +                             EVENT_CYCLES_COUNTER,
> > +                             false);
> > +       /*
> > +        * When the cycle counter overflows, all counters are stopped,
> > +        * and an IRQ is raised. If any other counter overflows, it
> > +        * continues counting, and no IRQ is raised.
> > +        *
> > +        * Cycles occur at least 4 times as often as other events, so we
> > +        * can update all events on a cycle counter overflow and not
> > +        * lose events.
> > +        *
> > +        */
> > +       for (i = 0; i < NUM_COUNTERS; i++) {
> > +
> > +               if (!pmu->events[i])
> > +                       continue;
> > +
> > +               event = pmu->events[i];
> > +
> > +               ddr_perf_event_update(event);
> > +
> > +               if (event->hw.idx == EVENT_CYCLES_COUNTER)
> > +                       cycle_event = event;
> > +       }
> > +
> > +       ddr_perf_counter_enable(pmu,
> > +                             EVENT_CYCLES_ID,
> > +                             EVENT_CYCLES_COUNTER,
> > +                             true);
> > +       if (cycle_event)
> > +               ddr_perf_event_update(cycle_event);
> > +
> > +       return IRQ_HANDLED;
> > +}
> > +
> > +static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
> > +{
> > +       struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
> > +       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;
> > +
> > +       WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));
> > +
> > +       return 0;
> > +}
> > +
> > +static int ddr_perf_probe(struct platform_device *pdev)
> > +{
> > +       struct ddr_pmu *pmu;
> > +       struct device_node *np;
> > +       void __iomem *base;
> > +       char *name;
> > +       int num;
> > +       int ret;
> > +       int irq;
> > +
> > +       base = devm_platform_ioremap_resource(pdev, 0);
> > +       if (IS_ERR(base))
> > +               return PTR_ERR(base);
> > +
> > +       np = pdev->dev.of_node;
> > +
> > +       pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
> > +       if (!pmu)
> > +               return -ENOMEM;
> > +
> > +       num = ddr_perf_init(pmu, base, &pdev->dev);
> > +
> > +       platform_set_drvdata(pdev, pmu);
> > +
> > +       name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "imx8_ddr%d", num);
> > +       if (!name)
> > +               return -ENOMEM;
> > +
> > +       pmu->cpu = raw_smp_processor_id();
> > +       ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> > +                                     DDR_PERF_DEV_NAME,
> > +                                     NULL,
> > +                                     ddr_perf_offline_cpu);
> > +
> > +       if (ret < 0) {
> > +               dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       pmu->cpuhp_state = ret;
> > +
> > +       /* Register the pmu instance for cpu hotplug */
> > +       cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +
> > +       /* Request irq */
> > +       irq = of_irq_get(np, 0);
> > +       if (irq < 0) {
> > +               dev_err(&pdev->dev, "Failed to get irq: %d", irq);
> > +               ret = irq;
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       ret = devm_request_irq(&pdev->dev, irq,
> > +                                       ddr_perf_irq_handler,
> > +                                       IRQF_NOBALANCING | IRQF_NO_THREAD,
> > +                                       DDR_PERF_DEV_NAME,
> > +                                       pmu);
> > +       if (ret < 0) {
> > +               dev_err(&pdev->dev, "Request irq failed: %d", ret);
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       pmu->irq = irq;
> > +       ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
> > +       if (ret) {
> > +               dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
> > +               goto ddr_perf_err;
> > +       }
> > +
> > +       ret = perf_pmu_register(&pmu->pmu, name, -1);
> > +       if (ret)
> > +               goto ddr_perf_err;
> > +
> > +       return 0;
> > +
> > +ddr_perf_err:
> > +       if (pmu->cpuhp_state)
> > +               cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +
> > +       ida_simple_remove(&ddr_ida, pmu->id);
> > +       dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
> > +       return ret;
> > +}
> > +
> > +static int ddr_perf_remove(struct platform_device *pdev)
> > +{
> > +       struct ddr_pmu *pmu = platform_get_drvdata(pdev);
> > +
> > +       cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
> > +       irq_set_affinity_hint(pmu->irq, NULL);
> > +
> > +       perf_pmu_unregister(&pmu->pmu);
> > +
> > +       ida_simple_remove(&ddr_ida, pmu->id);
> > +       return 0;
> > +}
> > +
> > +static struct platform_driver imx_ddr_pmu_driver = {
> > +       .driver         = {
> > +               .name   = "imx-ddr-pmu",
> > +               .of_match_table = imx_ddr_pmu_dt_ids,
> > +       },
> > +       .probe          = ddr_perf_probe,
> > +       .remove         = ddr_perf_remove,
> > +};
> > +
> > +module_platform_driver(imx_ddr_pmu_driver);
> > +MODULE_LICENSE("GPL v2");
> > --
> > 2.5.2
> >

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-05-14 13:53       ` Zhi Li
@ 2019-05-14 14:55         ` Will Deacon
  -1 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-05-14 14:55 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, Andrey Smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, May 14, 2019 at 08:53:18AM -0500, Zhi Li wrote:
> On Tue, May 7, 2019 at 2:30 PM Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
> >
> > On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
> > >
> > > Add ddr performance monitor support for iMX8QXP
> > >
> > > There are 4 counters for ddr perfomance events.
> > > counter 0 is dedicated for cycles.
> > > you choose any up to 3 no cycles events.
> > >
> > > for example:
> > >
> > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > >
> > > Support below events.
> > >
> > >   imx8_ddr0/activate/                                     [Kernel PMU event]
> > >   imx8_ddr0/axid-read/                                    [Kernel PMU event]
> > >   imx8_ddr0/axid-write/                                   [Kernel PMU event]
> > >   imx8_ddr0/cycles/                                       [Kernel PMU event]
> > >   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
> > >   imx8_ddr0/hp-read/                                      [Kernel PMU event]
> > >   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
> > >   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/load-mode/                                    [Kernel PMU event]
> > >   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
> > >   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
> > >   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/mwr/                                          [Kernel PMU event]
> > >   imx8_ddr0/precharge/                                    [Kernel PMU event]
> > >   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
> > >   imx8_ddr0/read-access/                                  [Kernel PMU event]
> > >   imx8_ddr0/read-activate/                                [Kernel PMU event]
> > >   imx8_ddr0/read-command/                                 [Kernel PMU event]
> > >   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
> > >   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
> > >   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
> > >   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
> > >   imx8_ddr0/read/                                         [Kernel PMU event]
> > >   imx8_ddr0/refresh/                                      [Kernel PMU event]
> > >   imx8_ddr0/selfresh/                                     [Kernel PMU event]
> > >   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/write-access/                                 [Kernel PMU event]
> > >   imx8_ddr0/write-command/                                [Kernel PMU event]
> > >   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
> > >   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
> > >   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
> > >   imx8_ddr0/write/
> > >
> > > Signed-off-by: Frank Li <Frank.Li@nxp.com>
> >
> > Don't really have anything to suggest, that I already haven't. LGTM, so:
> >
> > Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>
> 
> 
> Will Deacon:
>             Is it okay to pick up these patches?  Andrey always acked.

It's the merge window at the moment, so it's too late for 5.2. I'll have a
look at these in a couple of weeks as candidates for 5.3.

Thanks,

Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-05-14 14:55         ` Will Deacon
  0 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-05-14 14:55 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, Andrey Smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, May 14, 2019 at 08:53:18AM -0500, Zhi Li wrote:
> On Tue, May 7, 2019 at 2:30 PM Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
> >
> > On Wed, May 1, 2019 at 11:43 AM Frank Li <frank.li@nxp.com> wrote:
> > >
> > > Add ddr performance monitor support for iMX8QXP
> > >
> > > There are 4 counters for ddr perfomance events.
> > > counter 0 is dedicated for cycles.
> > > you choose any up to 3 no cycles events.
> > >
> > > for example:
> > >
> > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > >
> > > Support below events.
> > >
> > >   imx8_ddr0/activate/                                     [Kernel PMU event]
> > >   imx8_ddr0/axid-read/                                    [Kernel PMU event]
> > >   imx8_ddr0/axid-write/                                   [Kernel PMU event]
> > >   imx8_ddr0/cycles/                                       [Kernel PMU event]
> > >   imx8_ddr0/hp-read-credit-cnt/                           [Kernel PMU event]
> > >   imx8_ddr0/hp-read/                                      [Kernel PMU event]
> > >   imx8_ddr0/hp-req-nodcredit/                             [Kernel PMU event]
> > >   imx8_ddr0/hp-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/load-mode/                                    [Kernel PMU event]
> > >   imx8_ddr0/lp-read-credit-cnt/                           [Kernel PMU event]
> > >   imx8_ddr0/lp-req-nocredit/                              [Kernel PMU event]
> > >   imx8_ddr0/lp-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/mwr/                                          [Kernel PMU event]
> > >   imx8_ddr0/precharge/                                    [Kernel PMU event]
> > >   imx8_ddr0/raw-hazard/                                   [Kernel PMU event]
> > >   imx8_ddr0/read-access/                                  [Kernel PMU event]
> > >   imx8_ddr0/read-activate/                                [Kernel PMU event]
> > >   imx8_ddr0/read-command/                                 [Kernel PMU event]
> > >   imx8_ddr0/read-cycles/                                  [Kernel PMU event]
> > >   imx8_ddr0/read-modify-write-command/                    [Kernel PMU event]
> > >   imx8_ddr0/read-queue-depth/                             [Kernel PMU event]
> > >   imx8_ddr0/read-write-transition/                        [Kernel PMU event]
> > >   imx8_ddr0/read/                                         [Kernel PMU event]
> > >   imx8_ddr0/refresh/                                      [Kernel PMU event]
> > >   imx8_ddr0/selfresh/                                     [Kernel PMU event]
> > >   imx8_ddr0/wr-xact-credit/                               [Kernel PMU event]
> > >   imx8_ddr0/write-access/                                 [Kernel PMU event]
> > >   imx8_ddr0/write-command/                                [Kernel PMU event]
> > >   imx8_ddr0/write-credit-cnt/                             [Kernel PMU event]
> > >   imx8_ddr0/write-cycles/                                 [Kernel PMU event]
> > >   imx8_ddr0/write-queue-depth/                            [Kernel PMU event]
> > >   imx8_ddr0/write/
> > >
> > > Signed-off-by: Frank Li <Frank.Li@nxp.com>
> >
> > Don't really have anything to suggest, that I already haven't. LGTM, so:
> >
> > Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>
> 
> 
> Will Deacon:
>             Is it okay to pick up these patches?  Andrey always acked.

It's the merge window at the moment, so it's too late for 5.2. I'll have a
look at these in a couple of weeks as candidates for 5.3.

Thanks,

Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-05-01 18:43   ` Frank Li
@ 2019-06-13 11:23     ` Will Deacon
  -1 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-13 11:23 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, shawnguo,
	s.hauer, robh+dt, dl-linux-imx, kernel, lznuaa, festevam,
	linux-arm-kernel

On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> Add ddr performance monitor support for iMX8QXP
> 
> There are 4 counters for ddr perfomance events.
> counter 0 is dedicated for cycles.
> you choose any up to 3 no cycles events.
> 
> for example:
> 
> perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls

I've pushed patches 1, 2 and 4 out with some minor tweaks to:

https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf

I'll leave the actual .dts change to go via the soc tree, since last time
I took one of those it just resulted in conflicts.

Frank, Andrey: Please could you try to run the perf fuzzer on this before
it lands in mainline? It has a good track record of finding nasty PMU driver
bugs, but it obviously requires access to hardware which implements the PMU:

http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/

Cheers,

Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-13 11:23     ` Will Deacon
  0 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-13 11:23 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, shawnguo,
	s.hauer, robh+dt, dl-linux-imx, kernel, lznuaa, festevam,
	linux-arm-kernel

On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> Add ddr performance monitor support for iMX8QXP
> 
> There are 4 counters for ddr perfomance events.
> counter 0 is dedicated for cycles.
> you choose any up to 3 no cycles events.
> 
> for example:
> 
> perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls

I've pushed patches 1, 2 and 4 out with some minor tweaks to:

https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf

I'll leave the actual .dts change to go via the soc tree, since last time
I took one of those it just resulted in conflicts.

Frank, Andrey: Please could you try to run the perf fuzzer on this before
it lands in mainline? It has a good track record of finding nasty PMU driver
bugs, but it obviously requires access to hardware which implements the PMU:

http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/

Cheers,

Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-13 11:23     ` Will Deacon
@ 2019-06-13 17:04       ` Zhi Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-13 17:04 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
>
> On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > Add ddr performance monitor support for iMX8QXP
> >
> > There are 4 counters for ddr perfomance events.
> > counter 0 is dedicated for cycles.
> > you choose any up to 3 no cycles events.
> >
> > for example:
> >
> > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
>
> I've pushed patches 1, 2 and 4 out with some minor tweaks to:
>
> https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
>
> I'll leave the actual .dts change to go via the soc tree, since last time
> I took one of those it just resulted in conflicts.
>
> Frank, Andrey: Please could you try to run the perf fuzzer on this before
> it lands in mainline? It has a good track record of finding nasty PMU driver
> bugs, but it obviously requires access to hardware which implements the PMU:
>
> http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/

Okay, how long should be run generally?
I need make sure it can pass without my patches at our platform.

Best regards
Frank Li

>
> Cheers,
>
> Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-13 17:04       ` Zhi Li
  0 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-13 17:04 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
>
> On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > Add ddr performance monitor support for iMX8QXP
> >
> > There are 4 counters for ddr perfomance events.
> > counter 0 is dedicated for cycles.
> > you choose any up to 3 no cycles events.
> >
> > for example:
> >
> > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
>
> I've pushed patches 1, 2 and 4 out with some minor tweaks to:
>
> https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
>
> I'll leave the actual .dts change to go via the soc tree, since last time
> I took one of those it just resulted in conflicts.
>
> Frank, Andrey: Please could you try to run the perf fuzzer on this before
> it lands in mainline? It has a good track record of finding nasty PMU driver
> bugs, but it obviously requires access to hardware which implements the PMU:
>
> http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/

Okay, how long should be run generally?
I need make sure it can pass without my patches at our platform.

Best regards
Frank Li

>
> Cheers,
>
> Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-13 17:04       ` Zhi Li
@ 2019-06-13 17:44         ` Will Deacon
  -1 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-13 17:44 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > Add ddr performance monitor support for iMX8QXP
> > >
> > > There are 4 counters for ddr perfomance events.
> > > counter 0 is dedicated for cycles.
> > > you choose any up to 3 no cycles events.
> > >
> > > for example:
> > >
> > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> >
> > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> >
> > I'll leave the actual .dts change to go via the soc tree, since last time
> > I took one of those it just resulted in conflicts.
> >
> > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > it lands in mainline? It has a good track record of finding nasty PMU driver
> > bugs, but it obviously requires access to hardware which implements the PMU:
> >
> > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> 
> Okay, how long should be run generally?
> I need make sure it can pass without my patches at our platform.

As you long as you can really, but if it survives a few hours that's usually
a good sign. Overnight is even better.

Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-13 17:44         ` Will Deacon
  0 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-13 17:44 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > Add ddr performance monitor support for iMX8QXP
> > >
> > > There are 4 counters for ddr perfomance events.
> > > counter 0 is dedicated for cycles.
> > > you choose any up to 3 no cycles events.
> > >
> > > for example:
> > >
> > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> >
> > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> >
> > I'll leave the actual .dts change to go via the soc tree, since last time
> > I took one of those it just resulted in conflicts.
> >
> > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > it lands in mainline? It has a good track record of finding nasty PMU driver
> > bugs, but it obviously requires access to hardware which implements the PMU:
> >
> > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> 
> Okay, how long should be run generally?
> I need make sure it can pass without my patches at our platform.

As you long as you can really, but if it survives a few hours that's usually
a good sign. Overnight is even better.

Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-13 17:44         ` Will Deacon
@ 2019-06-13 19:13           ` Zhi Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-13 19:13 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
>
> On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > >
> > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > Add ddr performance monitor support for iMX8QXP
> > > >
> > > > There are 4 counters for ddr perfomance events.
> > > > counter 0 is dedicated for cycles.
> > > > you choose any up to 3 no cycles events.
> > > >
> > > > for example:
> > > >
> > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > >
> > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > >
> > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > I took one of those it just resulted in conflicts.
> > >
> > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > bugs, but it obviously requires access to hardware which implements the PMU:
> > >
> > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> >
> > Okay, how long should be run generally?
> > I need make sure it can pass without my patches at our platform.
>
> As you long as you can really, but if it survives a few hours that's usually
> a good sign. Overnight is even better.

Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Sun Jun 2 13:55:33 2019 -0700

    Linux 5.2-rc3

RCU report problem:

[ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
[ 6048.747550] rcu:     1-....: (5249 ticks this GP)
idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
[ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
[ 6048.761745] Task dump for CPU 1:
[ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
[ 6048.772030] Call trace:
[ 6048.774493]  dump_backtrace+0x0/0x130
[ 6048.778159]  show_stack+0x14/0x20
[ 6048.781477]  sched_show_task+0x108/0x138
[ 6048.785401]  dump_cpu_task+0x40/0x4c
[ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
[ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
[ 6048.797357]  update_process_times+0x2c/0x70
[ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
[ 6048.806076]  tick_sched_timer+0x48/0x98
[ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
[ 6048.814277]  hrtimer_interrupt+0xe4/0x238
[ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
[ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
[ 6048.827277]  generic_handle_irq+0x24/0x38

Only armv8_pmuv3 and basic perf event enabled.
perf list

List of pre-defined events (to be used in -e):

  branch-instructions OR branches                    [Hardware event]
  branch-misses                                      [Hardware event]
  bus-cycles                                         [Hardware event]
  cache-misses                                       [Hardware event]
  cache-references                                   [Hardware event]
  cpu-cycles OR cycles                               [Hardware event]
  instructions                                       [Hardware event]

  alignment-faults                                   [Software event]
  bpf-output                                         [Software event]
  context-switches OR cs                             [Software event]
  cpu-clock                                          [Software event]
  cpu-migrations OR migrations                       [Software event]
  dummy                                              [Software event]
  emulation-faults                                   [Software event]
  major-faults                                       [Software event]
  minor-faults                                       [Software event]
  page-faults OR faults                              [Software event]
  task-clock                                         [Software event]

  L1-dcache-load-misses                              [Hardware cache event]
  L1-dcache-loads                                    [Hardware cache event]
  L1-icache-load-misses                              [Hardware cache event]
  L1-icache-loads                                    [Hardware cache event]
  branch-load-misses                                 [Hardware cache event]
  branch-loads                                       [Hardware cache event]
  dTLB-load-misses                                   [Hardware cache event]
  iTLB-load-misses                                   [Hardware cache event]

  armv8_pmuv3/br_immed_retired/                      [Kernel PMU event]
  armv8_pmuv3/br_mis_pred/                           [Kernel PMU event]
 ****8

>
> Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-13 19:13           ` Zhi Li
  0 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-13 19:13 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
>
> On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > >
> > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > Add ddr performance monitor support for iMX8QXP
> > > >
> > > > There are 4 counters for ddr perfomance events.
> > > > counter 0 is dedicated for cycles.
> > > > you choose any up to 3 no cycles events.
> > > >
> > > > for example:
> > > >
> > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > >
> > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > >
> > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > I took one of those it just resulted in conflicts.
> > >
> > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > bugs, but it obviously requires access to hardware which implements the PMU:
> > >
> > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> >
> > Okay, how long should be run generally?
> > I need make sure it can pass without my patches at our platform.
>
> As you long as you can really, but if it survives a few hours that's usually
> a good sign. Overnight is even better.

Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Sun Jun 2 13:55:33 2019 -0700

    Linux 5.2-rc3

RCU report problem:

[ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
[ 6048.747550] rcu:     1-....: (5249 ticks this GP)
idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
[ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
[ 6048.761745] Task dump for CPU 1:
[ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
[ 6048.772030] Call trace:
[ 6048.774493]  dump_backtrace+0x0/0x130
[ 6048.778159]  show_stack+0x14/0x20
[ 6048.781477]  sched_show_task+0x108/0x138
[ 6048.785401]  dump_cpu_task+0x40/0x4c
[ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
[ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
[ 6048.797357]  update_process_times+0x2c/0x70
[ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
[ 6048.806076]  tick_sched_timer+0x48/0x98
[ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
[ 6048.814277]  hrtimer_interrupt+0xe4/0x238
[ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
[ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
[ 6048.827277]  generic_handle_irq+0x24/0x38

Only armv8_pmuv3 and basic perf event enabled.
perf list

List of pre-defined events (to be used in -e):

  branch-instructions OR branches                    [Hardware event]
  branch-misses                                      [Hardware event]
  bus-cycles                                         [Hardware event]
  cache-misses                                       [Hardware event]
  cache-references                                   [Hardware event]
  cpu-cycles OR cycles                               [Hardware event]
  instructions                                       [Hardware event]

  alignment-faults                                   [Software event]
  bpf-output                                         [Software event]
  context-switches OR cs                             [Software event]
  cpu-clock                                          [Software event]
  cpu-migrations OR migrations                       [Software event]
  dummy                                              [Software event]
  emulation-faults                                   [Software event]
  major-faults                                       [Software event]
  minor-faults                                       [Software event]
  page-faults OR faults                              [Software event]
  task-clock                                         [Software event]

  L1-dcache-load-misses                              [Hardware cache event]
  L1-dcache-loads                                    [Hardware cache event]
  L1-icache-load-misses                              [Hardware cache event]
  L1-icache-loads                                    [Hardware cache event]
  branch-load-misses                                 [Hardware cache event]
  branch-loads                                       [Hardware cache event]
  dTLB-load-misses                                   [Hardware cache event]
  iTLB-load-misses                                   [Hardware cache event]

  armv8_pmuv3/br_immed_retired/                      [Kernel PMU event]
  armv8_pmuv3/br_mis_pred/                           [Kernel PMU event]
 ****8

>
> Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-13 19:13           ` Zhi Li
@ 2019-06-14 10:23             ` Will Deacon
  -1 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-14 10:23 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > >
> > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > Add ddr performance monitor support for iMX8QXP
> > > > >
> > > > > There are 4 counters for ddr perfomance events.
> > > > > counter 0 is dedicated for cycles.
> > > > > you choose any up to 3 no cycles events.
> > > > >
> > > > > for example:
> > > > >
> > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > >
> > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > >
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > >
> > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > I took one of those it just resulted in conflicts.
> > > >
> > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > >
> > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > >
> > > Okay, how long should be run generally?
> > > I need make sure it can pass without my patches at our platform.
> >
> > As you long as you can really, but if it survives a few hours that's usually
> > a good sign. Overnight is even better.
> 
> Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> Author: Linus Torvalds <torvalds@linux-foundation.org>
> Date:   Sun Jun 2 13:55:33 2019 -0700
> 
>     Linux 5.2-rc3
> 
> RCU report problem:
> 
> [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> [ 6048.761745] Task dump for CPU 1:
> [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> [ 6048.772030] Call trace:
> [ 6048.774493]  dump_backtrace+0x0/0x130
> [ 6048.778159]  show_stack+0x14/0x20
> [ 6048.781477]  sched_show_task+0x108/0x138
> [ 6048.785401]  dump_cpu_task+0x40/0x4c
> [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> [ 6048.797357]  update_process_times+0x2c/0x70
> [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> [ 6048.806076]  tick_sched_timer+0x48/0x98
> [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> [ 6048.827277]  generic_handle_irq+0x24/0x38

This is the timer interrupt which prompts the RCU splat. Do you have
information about where the CPU was when the interrupt occurred?

In the meantime, it's still worth leaving the fuzzer running to see what
else it finds.

Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-14 10:23             ` Will Deacon
  0 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-14 10:23 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > >
> > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > Add ddr performance monitor support for iMX8QXP
> > > > >
> > > > > There are 4 counters for ddr perfomance events.
> > > > > counter 0 is dedicated for cycles.
> > > > > you choose any up to 3 no cycles events.
> > > > >
> > > > > for example:
> > > > >
> > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > >
> > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > >
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > >
> > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > I took one of those it just resulted in conflicts.
> > > >
> > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > >
> > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > >
> > > Okay, how long should be run generally?
> > > I need make sure it can pass without my patches at our platform.
> >
> > As you long as you can really, but if it survives a few hours that's usually
> > a good sign. Overnight is even better.
> 
> Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> Author: Linus Torvalds <torvalds@linux-foundation.org>
> Date:   Sun Jun 2 13:55:33 2019 -0700
> 
>     Linux 5.2-rc3
> 
> RCU report problem:
> 
> [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> [ 6048.761745] Task dump for CPU 1:
> [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> [ 6048.772030] Call trace:
> [ 6048.774493]  dump_backtrace+0x0/0x130
> [ 6048.778159]  show_stack+0x14/0x20
> [ 6048.781477]  sched_show_task+0x108/0x138
> [ 6048.785401]  dump_cpu_task+0x40/0x4c
> [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> [ 6048.797357]  update_process_times+0x2c/0x70
> [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> [ 6048.806076]  tick_sched_timer+0x48/0x98
> [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> [ 6048.827277]  generic_handle_irq+0x24/0x38

This is the timer interrupt which prompts the RCU splat. Do you have
information about where the CPU was when the interrupt occurred?

In the meantime, it's still worth leaving the fuzzer running to see what
else it finds.

Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 3/4] arm64: dts: imx8qxp: added ddr performance monitor nodes
  2019-05-01 18:43   ` Frank Li
@ 2019-06-18  6:37     ` Shawn Guo
  -1 siblings, 0 replies; 30+ messages in thread
From: Shawn Guo @ 2019-06-18  6:37 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, s.hauer,
	will.deacon, robh+dt, dl-linux-imx, kernel, lznuaa, festevam,
	linux-arm-kernel

On Wed, May 01, 2019 at 06:43:32PM +0000, Frank Li wrote:
> Add ddr performance monitor
> 
> Signed-off-by: Frank Li <Frank.Li@nxp.com>
> Reviewed-by: Fabio Estevam <festevam@gmail.com>

Please resend without using 'Content-Transfer-Encoding: base64'.

Shawn

> ---
> 
> Notes:
>     No change from v9 to v12
> 
>     Change from v8 to v9
>     * put ddr-pmu under ddr_subsystem bus
> 
>     Change from v3 to v8
>     * none
> 
>     Change from v2 to v3
>     * ddr_pmu0 -> ddr-pmu
> 
>  arch/arm64/boot/dts/freescale/imx8qxp.dtsi | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> index 0683ee2..a33e08c 100644
> --- a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> +++ b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> @@ -378,6 +378,20 @@
>  		};
>  	};
>  
> +	ddr_subsyss: bus@5c000000 {
> +		compatible = "simple-bus";
> +		#address-cells = <1>;
> +		#size-cells = <1>;
> +		ranges = <0x5c000000 0x0 0x5c000000 0x1000000>;
> +
> +		ddr-pmu@5c020000 {
> +			compatible = "fsl,imx8-ddr-pmu";
> +			reg = <0x5c020000 0x10000>;
> +			interrupt-parent = <&gic>;
> +			interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
> +		};
> +	};
> +
>  	lsio_subsys: bus@5d000000 {
>  		compatible = "simple-bus";
>  		#address-cells = <1>;
> -- 
> 2.5.2
> 

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 3/4] arm64: dts: imx8qxp: added ddr performance monitor nodes
@ 2019-06-18  6:37     ` Shawn Guo
  0 siblings, 0 replies; 30+ messages in thread
From: Shawn Guo @ 2019-06-18  6:37 UTC (permalink / raw)
  To: Frank Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, s.hauer,
	will.deacon, robh+dt, dl-linux-imx, kernel, lznuaa, festevam,
	linux-arm-kernel

On Wed, May 01, 2019 at 06:43:32PM +0000, Frank Li wrote:
> Add ddr performance monitor
> 
> Signed-off-by: Frank Li <Frank.Li@nxp.com>
> Reviewed-by: Fabio Estevam <festevam@gmail.com>

Please resend without using 'Content-Transfer-Encoding: base64'.

Shawn

> ---
> 
> Notes:
>     No change from v9 to v12
> 
>     Change from v8 to v9
>     * put ddr-pmu under ddr_subsystem bus
> 
>     Change from v3 to v8
>     * none
> 
>     Change from v2 to v3
>     * ddr_pmu0 -> ddr-pmu
> 
>  arch/arm64/boot/dts/freescale/imx8qxp.dtsi | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> index 0683ee2..a33e08c 100644
> --- a/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> +++ b/arch/arm64/boot/dts/freescale/imx8qxp.dtsi
> @@ -378,6 +378,20 @@
>  		};
>  	};
>  
> +	ddr_subsyss: bus@5c000000 {
> +		compatible = "simple-bus";
> +		#address-cells = <1>;
> +		#size-cells = <1>;
> +		ranges = <0x5c000000 0x0 0x5c000000 0x1000000>;
> +
> +		ddr-pmu@5c020000 {
> +			compatible = "fsl,imx8-ddr-pmu";
> +			reg = <0x5c020000 0x10000>;
> +			interrupt-parent = <&gic>;
> +			interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
> +		};
> +	};
> +
>  	lsio_subsys: bus@5d000000 {
>  		compatible = "simple-bus";
>  		#address-cells = <1>;
> -- 
> 2.5.2
> 

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-14 10:23             ` Will Deacon
@ 2019-06-18 17:24               ` Zhi Li
  -1 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-18 17:24 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Fri, Jun 14, 2019 at 5:23 AM Will Deacon <will.deacon@arm.com> wrote:
>
> On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> > On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> > >
> > > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > > >
> > > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > > Add ddr performance monitor support for iMX8QXP
> > > > > >
> > > > > > There are 4 counters for ddr perfomance events.
> > > > > > counter 0 is dedicated for cycles.
> > > > > > you choose any up to 3 no cycles events.
> > > > > >
> > > > > > for example:
> > > > > >
> > > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > > >
> > > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > > >
> > > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > > >
> > > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > > I took one of those it just resulted in conflicts.
> > > > >
> > > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > > >
> > > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > > >
> > > > Okay, how long should be run generally?
> > > > I need make sure it can pass without my patches at our platform.
> > >
> > > As you long as you can really, but if it survives a few hours that's usually
> > > a good sign. Overnight is even better.
> >
> > Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> > Author: Linus Torvalds <torvalds@linux-foundation.org>
> > Date:   Sun Jun 2 13:55:33 2019 -0700
> >
> >     Linux 5.2-rc3
> >
> > RCU report problem:
> >
> > [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> > [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> > idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> > [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> > [ 6048.761745] Task dump for CPU 1:
> > [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> > [ 6048.772030] Call trace:
> > [ 6048.774493]  dump_backtrace+0x0/0x130
> > [ 6048.778159]  show_stack+0x14/0x20
> > [ 6048.781477]  sched_show_task+0x108/0x138
> > [ 6048.785401]  dump_cpu_task+0x40/0x4c
> > [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> > [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> > [ 6048.797357]  update_process_times+0x2c/0x70
> > [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> > [ 6048.806076]  tick_sched_timer+0x48/0x98
> > [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> > [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> > [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> > [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> > [ 6048.827277]  generic_handle_irq+0x24/0x38
>
> This is the timer interrupt which prompts the RCU splat. Do you have
> information about where the CPU was when the interrupt occurred?
>
> In the meantime, it's still worth leaving the fuzzer running to see what
> else it finds.

Overnight test done, only above rcu problem happen at both with and
without ddr perf patches.

best regards
Frank Li


>
> Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-18 17:24               ` Zhi Li
  0 siblings, 0 replies; 30+ messages in thread
From: Zhi Li @ 2019-06-18 17:24 UTC (permalink / raw)
  To: Will Deacon
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Fri, Jun 14, 2019 at 5:23 AM Will Deacon <will.deacon@arm.com> wrote:
>
> On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> > On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> > >
> > > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > > >
> > > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > > Add ddr performance monitor support for iMX8QXP
> > > > > >
> > > > > > There are 4 counters for ddr perfomance events.
> > > > > > counter 0 is dedicated for cycles.
> > > > > > you choose any up to 3 no cycles events.
> > > > > >
> > > > > > for example:
> > > > > >
> > > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > > >
> > > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > > >
> > > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > > >
> > > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > > I took one of those it just resulted in conflicts.
> > > > >
> > > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > > >
> > > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > > >
> > > > Okay, how long should be run generally?
> > > > I need make sure it can pass without my patches at our platform.
> > >
> > > As you long as you can really, but if it survives a few hours that's usually
> > > a good sign. Overnight is even better.
> >
> > Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> > Author: Linus Torvalds <torvalds@linux-foundation.org>
> > Date:   Sun Jun 2 13:55:33 2019 -0700
> >
> >     Linux 5.2-rc3
> >
> > RCU report problem:
> >
> > [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> > [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> > idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> > [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> > [ 6048.761745] Task dump for CPU 1:
> > [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> > [ 6048.772030] Call trace:
> > [ 6048.774493]  dump_backtrace+0x0/0x130
> > [ 6048.778159]  show_stack+0x14/0x20
> > [ 6048.781477]  sched_show_task+0x108/0x138
> > [ 6048.785401]  dump_cpu_task+0x40/0x4c
> > [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> > [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> > [ 6048.797357]  update_process_times+0x2c/0x70
> > [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> > [ 6048.806076]  tick_sched_timer+0x48/0x98
> > [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> > [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> > [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> > [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> > [ 6048.827277]  generic_handle_irq+0x24/0x38
>
> This is the timer interrupt which prompts the RCU splat. Do you have
> information about where the CPU was when the interrupt occurred?
>
> In the meantime, it's still worth leaving the fuzzer running to see what
> else it finds.

Overnight test done, only above rcu problem happen at both with and
without ddr perf patches.

best regards
Frank Li


>
> Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
  2019-06-18 17:24               ` Zhi Li
@ 2019-06-18 17:58                 ` Will Deacon
  -1 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-18 17:58 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, Jun 18, 2019 at 12:24:19PM -0500, Zhi Li wrote:
> On Fri, Jun 14, 2019 at 5:23 AM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> > > On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> > > >
> > > > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > > > >
> > > > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > > > Add ddr performance monitor support for iMX8QXP
> > > > > > >
> > > > > > > There are 4 counters for ddr perfomance events.
> > > > > > > counter 0 is dedicated for cycles.
> > > > > > > you choose any up to 3 no cycles events.
> > > > > > >
> > > > > > > for example:
> > > > > > >
> > > > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > > > >
> > > > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > > > >
> > > > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > > > >
> > > > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > > > I took one of those it just resulted in conflicts.
> > > > > >
> > > > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > > > >
> > > > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > > > >
> > > > > Okay, how long should be run generally?
> > > > > I need make sure it can pass without my patches at our platform.
> > > >
> > > > As you long as you can really, but if it survives a few hours that's usually
> > > > a good sign. Overnight is even better.
> > >
> > > Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> > > Author: Linus Torvalds <torvalds@linux-foundation.org>
> > > Date:   Sun Jun 2 13:55:33 2019 -0700
> > >
> > >     Linux 5.2-rc3
> > >
> > > RCU report problem:
> > >
> > > [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> > > [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> > > idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> > > [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> > > [ 6048.761745] Task dump for CPU 1:
> > > [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> > > [ 6048.772030] Call trace:
> > > [ 6048.774493]  dump_backtrace+0x0/0x130
> > > [ 6048.778159]  show_stack+0x14/0x20
> > > [ 6048.781477]  sched_show_task+0x108/0x138
> > > [ 6048.785401]  dump_cpu_task+0x40/0x4c
> > > [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> > > [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> > > [ 6048.797357]  update_process_times+0x2c/0x70
> > > [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> > > [ 6048.806076]  tick_sched_timer+0x48/0x98
> > > [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> > > [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> > > [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> > > [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> > > [ 6048.827277]  generic_handle_irq+0x24/0x38
> >
> > This is the timer interrupt which prompts the RCU splat. Do you have
> > information about where the CPU was when the interrupt occurred?
> >
> > In the meantime, it's still worth leaving the fuzzer running to see what
> > else it finds.
> 
> Overnight test done, only above rcu problem happen at both with and
> without ddr perf patches.

Great, thanks for giving it a go.

Will

^ permalink raw reply	[flat|nested] 30+ messages in thread

* Re: [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support
@ 2019-06-18 17:58                 ` Will Deacon
  0 siblings, 0 replies; 30+ messages in thread
From: Will Deacon @ 2019-06-18 17:58 UTC (permalink / raw)
  To: Zhi Li
  Cc: mark.rutland, Aisheng Dong, devicetree, andrew.smirnov, festevam,
	s.hauer, Frank Li, robh+dt, dl-linux-imx, kernel, shawnguo,
	linux-arm-kernel

On Tue, Jun 18, 2019 at 12:24:19PM -0500, Zhi Li wrote:
> On Fri, Jun 14, 2019 at 5:23 AM Will Deacon <will.deacon@arm.com> wrote:
> >
> > On Thu, Jun 13, 2019 at 02:13:20PM -0500, Zhi Li wrote:
> > > On Thu, Jun 13, 2019 at 12:44 PM Will Deacon <will.deacon@arm.com> wrote:
> > > >
> > > > On Thu, Jun 13, 2019 at 12:04:37PM -0500, Zhi Li wrote:
> > > > > On Thu, Jun 13, 2019 at 6:23 AM Will Deacon <will.deacon@arm.com> wrote:
> > > > > >
> > > > > > On Wed, May 01, 2019 at 06:43:29PM +0000, Frank Li wrote:
> > > > > > > Add ddr performance monitor support for iMX8QXP
> > > > > > >
> > > > > > > There are 4 counters for ddr perfomance events.
> > > > > > > counter 0 is dedicated for cycles.
> > > > > > > you choose any up to 3 no cycles events.
> > > > > > >
> > > > > > > for example:
> > > > > > >
> > > > > > > perf stat -a -e imx8_ddr0/read-cycles/,imx8_ddr0/write-cycles/,imx8_ddr0/precharge/ ls
> > > > > > > perf stat -a -e imx8_ddr0/cycles/,imx8_ddr0/read-access/,imx8_ddr0/write-access/ ls
> > > > > >
> > > > > > I've pushed patches 1, 2 and 4 out with some minor tweaks to:
> > > > > >
> > > > > > https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-next/perf
> > > > > >
> > > > > > I'll leave the actual .dts change to go via the soc tree, since last time
> > > > > > I took one of those it just resulted in conflicts.
> > > > > >
> > > > > > Frank, Andrey: Please could you try to run the perf fuzzer on this before
> > > > > > it lands in mainline? It has a good track record of finding nasty PMU driver
> > > > > > bugs, but it obviously requires access to hardware which implements the PMU:
> > > > > >
> > > > > > http://web.eece.maine.edu/~vweaver/projects/perf_events/fuzzer/
> > > > >
> > > > > Okay, how long should be run generally?
> > > > > I need make sure it can pass without my patches at our platform.
> > > >
> > > > As you long as you can really, but if it survives a few hours that's usually
> > > > a good sign. Overnight is even better.
> > >
> > > Base on commit f2c7c76c5d0a443053e94adb9f0918fa2fb85c3a
> > > Author: Linus Torvalds <torvalds@linux-foundation.org>
> > > Date:   Sun Jun 2 13:55:33 2019 -0700
> > >
> > >     Linux 5.2-rc3
> > >
> > > RCU report problem:
> > >
> > > [ 6048.741784] rcu: INFO: rcu_preempt self-detected stall on CPU
> > > [ 6048.747550] rcu:     1-....: (5249 ticks this GP)
> > > idle=c5a/1/0x4000000000000004 softirq=503121/503121 fqs=2425
> > > [ 6048.757384]  (t=5253 jiffies g=1416105 q=117)
> > > [ 6048.761745] Task dump for CPU 1:
> > > [ 6048.764977] perf_fuzzer     R  running task        0 32520    426 0x00000202
> > > [ 6048.772030] Call trace:
> > > [ 6048.774493]  dump_backtrace+0x0/0x130
> > > [ 6048.778159]  show_stack+0x14/0x20
> > > [ 6048.781477]  sched_show_task+0x108/0x138
> > > [ 6048.785401]  dump_cpu_task+0x40/0x4c
> > > [ 6048.788983]  rcu_dump_cpu_stacks+0x94/0xd0
> > > [ 6048.793082]  rcu_sched_clock_irq+0x5e0/0x918
> > > [ 6048.797357]  update_process_times+0x2c/0x70
> > > [ 6048.801545]  tick_sched_handle.isra.6+0x3c/0x50
> > > [ 6048.806076]  tick_sched_timer+0x48/0x98
> > > [ 6048.809918]  __hrtimer_run_queues+0x118/0x1a8
> > > [ 6048.814277]  hrtimer_interrupt+0xe4/0x238
> > > [ 6048.818296]  arch_timer_handler_phys+0x2c/0x38
> > > [ 6048.822743]  handle_percpu_devid_irq+0x80/0x140
> > > [ 6048.827277]  generic_handle_irq+0x24/0x38
> >
> > This is the timer interrupt which prompts the RCU splat. Do you have
> > information about where the CPU was when the interrupt occurred?
> >
> > In the meantime, it's still worth leaving the fuzzer running to see what
> > else it finds.
> 
> Overnight test done, only above rcu problem happen at both with and
> without ddr perf patches.

Great, thanks for giving it a go.

Will

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 30+ messages in thread

end of thread, other threads:[~2019-06-18 17:58 UTC | newest]

Thread overview: 30+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-05-01 18:43 [PATCH V12 1/4] dt-bindings: perf: imx8-ddr: add imx8qxp ddr performance monitor Frank Li
2019-05-01 18:43 ` Frank Li
2019-05-01 18:43 ` [PATCH V12 2/4] drivers/perf: imx_ddr: Add ddr performance counter support Frank Li
2019-05-01 18:43   ` Frank Li
2019-05-07 19:30   ` Andrey Smirnov
2019-05-07 19:30     ` Andrey Smirnov
2019-05-14 13:53     ` Zhi Li
2019-05-14 13:53       ` Zhi Li
2019-05-14 14:55       ` Will Deacon
2019-05-14 14:55         ` Will Deacon
2019-06-13 11:23   ` Will Deacon
2019-06-13 11:23     ` Will Deacon
2019-06-13 17:04     ` Zhi Li
2019-06-13 17:04       ` Zhi Li
2019-06-13 17:44       ` Will Deacon
2019-06-13 17:44         ` Will Deacon
2019-06-13 19:13         ` Zhi Li
2019-06-13 19:13           ` Zhi Li
2019-06-14 10:23           ` Will Deacon
2019-06-14 10:23             ` Will Deacon
2019-06-18 17:24             ` Zhi Li
2019-06-18 17:24               ` Zhi Li
2019-06-18 17:58               ` Will Deacon
2019-06-18 17:58                 ` Will Deacon
2019-05-01 18:43 ` [PATCH V12 3/4] arm64: dts: imx8qxp: added ddr performance monitor nodes Frank Li
2019-05-01 18:43   ` Frank Li
2019-06-18  6:37   ` Shawn Guo
2019-06-18  6:37     ` Shawn Guo
2019-05-01 18:43 ` [PATCH V12 4/4] MAINTAINERS: Added imx DDR performonitor driver maintainer information Frank Li
2019-05-01 18:43   ` Frank Li

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.