Merge tag 'xgene-drivers-for-4.9' of https://github.com/AppliedMicro/xgene-next into next/drivers

Pull "X-Gene driver changes queued for v4.9" from Duc Dang:

This patch set includes:
+ X-Gene SoC Performance Monitoring Unit (PMU) driver

* tag 'xgene-drivers-for-4.9' of https://github.com/AppliedMicro/xgene-next:
  perf: xgene: Add APM X-Gene SoC Performance Monitoring Unit driver
  Documentation: Add documentation for APM X-Gene SoC PMU DTS binding
  MAINTAINERS: Add entry for APM X-Gene SoC PMU driver

6 files changed, 1573 insertions, 0 deletions

diff --git a/Documentation/devicetree/bindings/perf/apm-xgene-pmu.txt b/Documentation/devicetree/bindings/perf/apm-xgene-pmu.txt
new file mode 100644
index 000000000000..afb11cf693c0
--- /dev/null
+++ b/Documentation/devicetree/bindings/perf/apm-xgene-pmu.txt
@@ -0,0 +1,112 @@
+* APM X-Gene SoC PMU bindings
+
+This is APM X-Gene SoC PMU (Performance Monitoring Unit) module.
+The following PMU devices are supported:
+
+  L3C			- L3 cache controller
+  IOB			- IO bridge
+  MCB			- Memory controller bridge
+  MC			- Memory controller
+
+The following section describes the SoC PMU DT node binding.
+
+Required properties:
+- compatible		: Shall be "apm,xgene-pmu" for revision 1 or
+                          "apm,xgene-pmu-v2" for revision 2.
+- regmap-csw		: Regmap of the CPU switch fabric (CSW) resource.
+- regmap-mcba		: Regmap of the MCB-A (memory bridge) resource.
+- regmap-mcbb		: Regmap of the MCB-B (memory bridge) resource.
+- reg			: First resource shall be the CPU bus PMU resource.
+- interrupts            : Interrupt-specifier for PMU IRQ.
+
+Required properties for L3C subnode:
+- compatible		: Shall be "apm,xgene-pmu-l3c".
+- reg			: First resource shall be the L3C PMU resource.
+
+Required properties for IOB subnode:
+- compatible		: Shall be "apm,xgene-pmu-iob".
+- reg			: First resource shall be the IOB PMU resource.
+
+Required properties for MCB subnode:
+- compatible		: Shall be "apm,xgene-pmu-mcb".
+- reg			: First resource shall be the MCB PMU resource.
+- enable-bit-index	: The bit indicates if the according MCB is enabled.
+
+Required properties for MC subnode:
+- compatible		: Shall be "apm,xgene-pmu-mc".
+- reg			: First resource shall be the MC PMU resource.
+- enable-bit-index	: The bit indicates if the according MC is enabled.
+
+Example:
+	csw: csw@7e200000 {
+		compatible = "apm,xgene-csw", "syscon";
+		reg = <0x0 0x7e200000 0x0 0x1000>;
+	};
+
+	mcba: mcba@7e700000 {
+		compatible = "apm,xgene-mcb", "syscon";
+		reg = <0x0 0x7e700000 0x0 0x1000>;
+	};
+
+	mcbb: mcbb@7e720000 {
+		compatible = "apm,xgene-mcb", "syscon";
+		reg = <0x0 0x7e720000 0x0 0x1000>;
+	};
+
+	pmu: pmu@78810000 {
+		compatible = "apm,xgene-pmu-v2";
+		#address-cells = <2>;
+		#size-cells = <2>;
+		ranges;
+		regmap-csw = <&csw>;
+		regmap-mcba = <&mcba>;
+		regmap-mcbb = <&mcbb>;
+		reg = <0x0 0x78810000 0x0 0x1000>;
+		interrupts = <0x0 0x22 0x4>;
+
+		pmul3c@7e610000 {
+			compatible = "apm,xgene-pmu-l3c";
+			reg = <0x0 0x7e610000 0x0 0x1000>;
+		};
+
+		pmuiob@7e940000 {
+			compatible = "apm,xgene-pmu-iob";
+			reg = <0x0 0x7e940000 0x0 0x1000>;
+		};
+
+		pmucmcb@7e710000 {
+			compatible = "apm,xgene-pmu-mcb";
+			reg = <0x0 0x7e710000 0x0 0x1000>;
+			enable-bit-index = <0>;
+		};
+
+		pmucmcb@7e730000 {
+			compatible = "apm,xgene-pmu-mcb";
+			reg = <0x0 0x7e730000 0x0 0x1000>;
+			enable-bit-index = <1>;
+		};
+
+		pmucmc@7e810000 {
+			compatible = "apm,xgene-pmu-mc";
+			reg = <0x0 0x7e810000 0x0 0x1000>;
+			enable-bit-index = <0>;
+		};
+
+		pmucmc@7e850000 {
+			compatible = "apm,xgene-pmu-mc";
+			reg = <0x0 0x7e850000 0x0 0x1000>;
+			enable-bit-index = <1>;
+		};
+
+		pmucmc@7e890000 {
+			compatible = "apm,xgene-pmu-mc";
+			reg = <0x0 0x7e890000 0x0 0x1000>;
+			enable-bit-index = <2>;
+		};
+
+		pmucmc@7e8d0000 {
+			compatible = "apm,xgene-pmu-mc";
+			reg = <0x0 0x7e8d0000 0x0 0x1000>;
+			enable-bit-index = <3>;
+		};
+	};
diff --git a/Documentation/perf/xgene-pmu.txt b/Documentation/perf/xgene-pmu.txt
new file mode 100644
index 000000000000..d7cff4454e5b
--- /dev/null
+++ b/Documentation/perf/xgene-pmu.txt
@@ -0,0 +1,48 @@
+APM X-Gene SoC Performance Monitoring Unit (PMU)
+================================================
+
+X-Gene SoC PMU consists of various independent system device PMUs such as
+L3 cache(s), I/O bridge(s), memory controller bridge(s) and memory
+controller(s). These PMU devices are loosely architected to follow the
+same model as the PMU for ARM cores. The PMUs share the same top level
+interrupt and status CSR region.
+
+PMU (perf) driver
+-----------------
+
+The xgene-pmu driver registers several perf PMU drivers. Each of the perf
+driver provides description of its available events and configuration options
+in sysfs, see /sys/devices/<l3cX/iobX/mcbX/mcX>/.
+
+The "format" directory describes format of the config (event ID),
+config1 (agent ID) fields of the perf_event_attr structure. The "events"
+directory provides configuration templates for all supported event types that
+can be used with perf tool. For example, "l3c0/bank-fifo-full/" is an
+equivalent of "l3c0/config=0x0b/".
+
+Most of the SoC PMU has a specific list of agent ID used for monitoring
+performance of a specific datapath. For example, agents of a L3 cache can be
+a specific CPU or an I/O bridge. Each PMU has a set of 2 registers capable of
+masking the agents from which the request come from. If the bit with
+the bit number corresponding to the agent is set, the event is counted only if
+it is caused by a request from that agent. Each agent ID bit is inversely mapped
+to a corresponding bit in "config1" field. By default, the event will be
+counted for all agent requests (config1 = 0x0). For all the supported agents of
+each PMU, please refer to APM X-Gene User Manual.
+
+Each perf driver also provides a "cpumask" sysfs attribute, which contains a
+single CPU ID of the processor which will be used to handle all the PMU events.
+
+Example for perf tool use:
+
+ / # perf list | grep -e l3c -e iob -e mcb -e mc
+   l3c0/ackq-full/                                    [Kernel PMU event]
+ <...>
+   mcb1/mcb-csw-stall/                                [Kernel PMU event]
+
+ / # perf stat -a -e l3c0/read-miss/,mcb1/csw-write-request/ sleep 1
+
+ / # perf stat -a -e l3c0/read-miss,config1=0xfffffffffffffffe/ sleep 1
+
+The driver does not support sampling, therefore "perf record" will
+not work. Per-task (without "-a") perf sessions are not supported.
diff --git a/MAINTAINERS b/MAINTAINERS
index 1cb3ce2db23c..2c207feda6d1 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -856,6 +856,13 @@ F:	drivers/net/phy/mdio-xgene.c
 F:	Documentation/devicetree/bindings/net/apm-xgene-enet.txt
 F:	Documentation/devicetree/bindings/net/apm-xgene-mdio.txt
 
+APPLIED MICRO (APM) X-GENE SOC PMU
+M:	Tai Nguyen <ttnguyen@apm.com>
+S:	Supported
+F:	drivers/perf/xgene_pmu.c
+F:	Documentation/perf/xgene-pmu.txt
+F:	Documentation/devicetree/bindings/perf/apm-xgene-pmu.txt
+
 APTINA CAMERA SENSOR PLL
 M:	Laurent Pinchart <Laurent.pinchart@ideasonboard.com>
 L:	linux-media@vger.kernel.org
diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index 04e2653bb8c0..4d5c5f9f0dbd 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -12,4 +12,11 @@ config ARM_PMU
 	  Say y if you want to use CPU performance monitors on ARM-based
 	  systems.
 
+config XGENE_PMU
+        depends on PERF_EVENTS && ARCH_XGENE
+        bool "APM X-Gene SoC PMU"
+        default n
+        help
+          Say y if you want to use APM X-Gene SoC performance monitors.
+
 endmenu
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index acd2397ded94..b116e982810b 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -1 +1,2 @@
 obj-$(CONFIG_ARM_PMU) += arm_pmu.o
+obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o
diff --git a/drivers/perf/xgene_pmu.c b/drivers/perf/xgene_pmu.c
new file mode 100644
index 000000000000..c2ac7646b99f
--- /dev/null
+++ b/drivers/perf/xgene_pmu.c
@@ -0,0 +1,1398 @@
+/*
+ * APM X-Gene SoC PMU (Performance Monitor Unit)
+ *
+ * Copyright (c) 2016, Applied Micro Circuits Corporation
+ * Author: Hoan Tran <hotran@apm.com>
+ *         Tai Nguyen <ttnguyen@apm.com>
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/acpi.h>
+#include <linux/clk.h>
+#include <linux/cpumask.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/mfd/syscon.h>
+#include <linux/of_address.h>
+#include <linux/of_fdt.h>
+#include <linux/of_irq.h>
+#include <linux/of_platform.h>
+#include <linux/perf_event.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+#include <linux/slab.h>
+
+#define CSW_CSWCR                       0x0000
+#define  CSW_CSWCR_DUALMCB_MASK         BIT(0)
+#define MCBADDRMR                       0x0000
+#define  MCBADDRMR_DUALMCU_MODE_MASK    BIT(2)
+
+#define PCPPMU_INTSTATUS_REG	0x000
+#define PCPPMU_INTMASK_REG	0x004
+#define  PCPPMU_INTMASK		0x0000000F
+#define  PCPPMU_INTENMASK	0xFFFFFFFF
+#define  PCPPMU_INTCLRMASK	0xFFFFFFF0
+#define  PCPPMU_INT_MCU		BIT(0)
+#define  PCPPMU_INT_MCB		BIT(1)
+#define  PCPPMU_INT_L3C		BIT(2)
+#define  PCPPMU_INT_IOB		BIT(3)
+
+#define PMU_MAX_COUNTERS	4
+#define PMU_CNT_MAX_PERIOD	0x100000000ULL
+#define PMU_OVERFLOW_MASK	0xF
+#define PMU_PMCR_E		BIT(0)
+#define PMU_PMCR_P		BIT(1)
+
+#define PMU_PMEVCNTR0		0x000
+#define PMU_PMEVCNTR1		0x004
+#define PMU_PMEVCNTR2		0x008
+#define PMU_PMEVCNTR3		0x00C
+#define PMU_PMEVTYPER0		0x400
+#define PMU_PMEVTYPER1		0x404
+#define PMU_PMEVTYPER2		0x408
+#define PMU_PMEVTYPER3		0x40C
+#define PMU_PMAMR0		0xA00
+#define PMU_PMAMR1		0xA04
+#define PMU_PMCNTENSET		0xC00
+#define PMU_PMCNTENCLR		0xC20
+#define PMU_PMINTENSET		0xC40
+#define PMU_PMINTENCLR		0xC60
+#define PMU_PMOVSR		0xC80
+#define PMU_PMCR		0xE04
+
+#define to_pmu_dev(p)     container_of(p, struct xgene_pmu_dev, pmu)
+#define GET_CNTR(ev)      (ev->hw.idx)
+#define GET_EVENTID(ev)   (ev->hw.config & 0xFFULL)
+#define GET_AGENTID(ev)   (ev->hw.config_base & 0xFFFFFFFFUL)
+#define GET_AGENT1ID(ev)  ((ev->hw.config_base >> 32) & 0xFFFFFFFFUL)
+
+struct hw_pmu_info {
+	u32 type;
+	u32 enable_mask;
+	void __iomem *csr;
+};
+
+struct xgene_pmu_dev {
+	struct hw_pmu_info *inf;
+	struct xgene_pmu *parent;
+	struct pmu pmu;
+	u8 max_counters;
+	DECLARE_BITMAP(cntr_assign_mask, PMU_MAX_COUNTERS);
+	u64 max_period;
+	const struct attribute_group **attr_groups;
+	struct perf_event *pmu_counter_event[PMU_MAX_COUNTERS];
+};
+
+struct xgene_pmu {
+	struct device *dev;
+	int version;
+	void __iomem *pcppmu_csr;
+	u32 mcb_active_mask;
+	u32 mc_active_mask;
+	cpumask_t cpu;
+	raw_spinlock_t lock;
+	struct list_head l3cpmus;
+	struct list_head iobpmus;
+	struct list_head mcbpmus;
+	struct list_head mcpmus;
+};
+
+struct xgene_pmu_dev_ctx {
+	char *name;
+	struct list_head next;
+	struct xgene_pmu_dev *pmu_dev;
+	struct hw_pmu_info inf;
+};
+
+struct xgene_pmu_data {
+	int id;
+	u32 data;
+};
+
+enum xgene_pmu_version {
+	PCP_PMU_V1 = 1,
+	PCP_PMU_V2,
+};
+
+enum xgene_pmu_dev_type {
+	PMU_TYPE_L3C = 0,
+	PMU_TYPE_IOB,
+	PMU_TYPE_MCB,
+	PMU_TYPE_MC,
+};
+
+/*
+ * sysfs format attributes
+ */
+static ssize_t xgene_pmu_format_show(struct device *dev,
+				     struct device_attribute *attr, char *buf)
+{
+	struct dev_ext_attribute *eattr;
+
+	eattr = container_of(attr, struct dev_ext_attribute, attr);
+	return sprintf(buf, "%s\n", (char *) eattr->var);
+}
+
+#define XGENE_PMU_FORMAT_ATTR(_name, _config)		\
+	(&((struct dev_ext_attribute[]) {		\
+		{ .attr = __ATTR(_name, S_IRUGO, xgene_pmu_format_show, NULL), \
+		  .var = (void *) _config, }		\
+	})[0].attr.attr)
+
+static struct attribute *l3c_pmu_format_attrs[] = {
+	XGENE_PMU_FORMAT_ATTR(l3c_eventid, "config:0-7"),
+	XGENE_PMU_FORMAT_ATTR(l3c_agentid, "config1:0-9"),
+	NULL,
+};
+
+static struct attribute *iob_pmu_format_attrs[] = {
+	XGENE_PMU_FORMAT_ATTR(iob_eventid, "config:0-7"),
+	XGENE_PMU_FORMAT_ATTR(iob_agentid, "config1:0-63"),
+	NULL,
+};
+
+static struct attribute *mcb_pmu_format_attrs[] = {
+	XGENE_PMU_FORMAT_ATTR(mcb_eventid, "config:0-5"),
+	XGENE_PMU_FORMAT_ATTR(mcb_agentid, "config1:0-9"),
+	NULL,
+};
+
+static struct attribute *mc_pmu_format_attrs[] = {
+	XGENE_PMU_FORMAT_ATTR(mc_eventid, "config:0-28"),
+	NULL,
+};
+
+static const struct attribute_group l3c_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = l3c_pmu_format_attrs,
+};
+
+static const struct attribute_group iob_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = iob_pmu_format_attrs,
+};
+
+static const struct attribute_group mcb_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = mcb_pmu_format_attrs,
+};
+
+static const struct attribute_group mc_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = mc_pmu_format_attrs,
+};
+
+/*
+ * sysfs event attributes
+ */
+static ssize_t xgene_pmu_event_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	struct dev_ext_attribute *eattr;
+
+	eattr = container_of(attr, struct dev_ext_attribute, attr);
+	return sprintf(buf, "config=0x%lx\n", (unsigned long) eattr->var);
+}
+
+#define XGENE_PMU_EVENT_ATTR(_name, _config)		\
+	(&((struct dev_ext_attribute[]) {		\
+		{ .attr = __ATTR(_name, S_IRUGO, xgene_pmu_event_show, NULL), \
+		  .var = (void *) _config, }		\
+	 })[0].attr.attr)
+
+static struct attribute *l3c_pmu_events_attrs[] = {
+	XGENE_PMU_EVENT_ATTR(cycle-count,			0x00),
+	XGENE_PMU_EVENT_ATTR(cycle-count-div-64,		0x01),
+	XGENE_PMU_EVENT_ATTR(read-hit,				0x02),
+	XGENE_PMU_EVENT_ATTR(read-miss,				0x03),
+	XGENE_PMU_EVENT_ATTR(write-need-replacement,		0x06),
+	XGENE_PMU_EVENT_ATTR(write-not-need-replacement,	0x07),
+	XGENE_PMU_EVENT_ATTR(tq-full,				0x08),
+	XGENE_PMU_EVENT_ATTR(ackq-full,				0x09),
+	XGENE_PMU_EVENT_ATTR(wdb-full,				0x0a),
+	XGENE_PMU_EVENT_ATTR(bank-fifo-full,			0x0b),
+	XGENE_PMU_EVENT_ATTR(odb-full,				0x0c),
+	XGENE_PMU_EVENT_ATTR(wbq-full,				0x0d),
+	XGENE_PMU_EVENT_ATTR(bank-conflict-fifo-issue,		0x0e),
+	XGENE_PMU_EVENT_ATTR(bank-fifo-issue,			0x0f),
+	NULL,
+};
+
+static struct attribute *iob_pmu_events_attrs[] = {
+	XGENE_PMU_EVENT_ATTR(cycle-count,			0x00),
+	XGENE_PMU_EVENT_ATTR(cycle-count-div-64,		0x01),
+	XGENE_PMU_EVENT_ATTR(axi0-read,				0x02),
+	XGENE_PMU_EVENT_ATTR(axi0-read-partial,			0x03),
+	XGENE_PMU_EVENT_ATTR(axi1-read,				0x04),
+	XGENE_PMU_EVENT_ATTR(axi1-read-partial,			0x05),
+	XGENE_PMU_EVENT_ATTR(csw-read-block,			0x06),
+	XGENE_PMU_EVENT_ATTR(csw-read-partial,			0x07),
+	XGENE_PMU_EVENT_ATTR(axi0-write,			0x10),
+	XGENE_PMU_EVENT_ATTR(axi0-write-partial,		0x11),
+	XGENE_PMU_EVENT_ATTR(axi1-write,			0x13),
+	XGENE_PMU_EVENT_ATTR(axi1-write-partial,		0x14),
+	XGENE_PMU_EVENT_ATTR(csw-inbound-dirty,			0x16),
+	NULL,
+};
+
+static struct attribute *mcb_pmu_events_attrs[] = {
+	XGENE_PMU_EVENT_ATTR(cycle-count,			0x00),
+	XGENE_PMU_EVENT_ATTR(cycle-count-div-64,		0x01),
+	XGENE_PMU_EVENT_ATTR(csw-read,				0x02),
+	XGENE_PMU_EVENT_ATTR(csw-write-request,			0x03),
+	XGENE_PMU_EVENT_ATTR(mcb-csw-stall,			0x04),
+	XGENE_PMU_EVENT_ATTR(cancel-read-gack,			0x05),
+	NULL,
+};
+
+static struct attribute *mc_pmu_events_attrs[] = {
+	XGENE_PMU_EVENT_ATTR(cycle-count,			0x00),
+	XGENE_PMU_EVENT_ATTR(cycle-count-div-64,		0x01),
+	XGENE_PMU_EVENT_ATTR(act-cmd-sent,			0x02),
+	XGENE_PMU_EVENT_ATTR(pre-cmd-sent,			0x03),
+	XGENE_PMU_EVENT_ATTR(rd-cmd-sent,			0x04),
+	XGENE_PMU_EVENT_ATTR(rda-cmd-sent,			0x05),
+	XGENE_PMU_EVENT_ATTR(wr-cmd-sent,			0x06),
+	XGENE_PMU_EVENT_ATTR(wra-cmd-sent,			0x07),
+	XGENE_PMU_EVENT_ATTR(pde-cmd-sent,			0x08),
+	XGENE_PMU_EVENT_ATTR(sre-cmd-sent,			0x09),
+	XGENE_PMU_EVENT_ATTR(prea-cmd-sent,			0x0a),
+	XGENE_PMU_EVENT_ATTR(ref-cmd-sent,			0x0b),
+	XGENE_PMU_EVENT_ATTR(rd-rda-cmd-sent,			0x0c),
+	XGENE_PMU_EVENT_ATTR(wr-wra-cmd-sent,			0x0d),
+	XGENE_PMU_EVENT_ATTR(in-rd-collision,			0x0e),
+	XGENE_PMU_EVENT_ATTR(in-wr-collision,			0x0f),
+	XGENE_PMU_EVENT_ATTR(collision-queue-not-empty,		0x10),
+	XGENE_PMU_EVENT_ATTR(collision-queue-full,		0x11),
+	XGENE_PMU_EVENT_ATTR(mcu-request,			0x12),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-request,			0x13),
+	XGENE_PMU_EVENT_ATTR(mcu-hp-rd-request,			0x14),
+	XGENE_PMU_EVENT_ATTR(mcu-wr-request,			0x15),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-proceed-all,		0x16),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-proceed-cancel,		0x17),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-response,			0x18),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-proceed-speculative-all,	0x19),
+	XGENE_PMU_EVENT_ATTR(mcu-rd-proceed-speculative-cancel,	0x1a),
+	XGENE_PMU_EVENT_ATTR(mcu-wr-proceed-all,		0x1b),
+	XGENE_PMU_EVENT_ATTR(mcu-wr-proceed-cancel,		0x1c),
+	NULL,
+};
+
+static const struct attribute_group l3c_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = l3c_pmu_events_attrs,
+};
+
+static const struct attribute_group iob_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = iob_pmu_events_attrs,
+};
+
+static const struct attribute_group mcb_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = mcb_pmu_events_attrs,
+};
+
+static const struct attribute_group mc_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = mc_pmu_events_attrs,
+};
+
+/*
+ * sysfs cpumask attributes
+ */
+static ssize_t xgene_pmu_cpumask_show(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(dev_get_drvdata(dev));
+
+	return cpumap_print_to_pagebuf(true, buf, &pmu_dev->parent->cpu);
+}
+
+static DEVICE_ATTR(cpumask, S_IRUGO, xgene_pmu_cpumask_show, NULL);
+
+static struct attribute *xgene_pmu_cpumask_attrs[] = {
+	&dev_attr_cpumask.attr,
+	NULL,
+};
+
+static const struct attribute_group pmu_cpumask_attr_group = {
+	.attrs = xgene_pmu_cpumask_attrs,
+};
+
+/*
+ * Per PMU device attribute groups
+ */
+static const struct attribute_group *l3c_pmu_attr_groups[] = {
+	&l3c_pmu_format_attr_group,
+	&pmu_cpumask_attr_group,
+	&l3c_pmu_events_attr_group,
+	NULL
+};
+
+static const struct attribute_group *iob_pmu_attr_groups[] = {
+	&iob_pmu_format_attr_group,
+	&pmu_cpumask_attr_group,
+	&iob_pmu_events_attr_group,
+	NULL
+};
+
+static const struct attribute_group *mcb_pmu_attr_groups[] = {
+	&mcb_pmu_format_attr_group,
+	&pmu_cpumask_attr_group,
+	&mcb_pmu_events_attr_group,
+	NULL
+};
+
+static const struct attribute_group *mc_pmu_attr_groups[] = {
+	&mc_pmu_format_attr_group,
+	&pmu_cpumask_attr_group,
+	&mc_pmu_events_attr_group,
+	NULL
+};
+
+static int get_next_avail_cntr(struct xgene_pmu_dev *pmu_dev)
+{
+	int cntr;
+
+	cntr = find_first_zero_bit(pmu_dev->cntr_assign_mask,
+				pmu_dev->max_counters);
+	if (cntr == pmu_dev->max_counters)
+		return -ENOSPC;
+	set_bit(cntr, pmu_dev->cntr_assign_mask);
+
+	return cntr;
+}
+
+static void clear_avail_cntr(struct xgene_pmu_dev *pmu_dev, int cntr)
+{
+	clear_bit(cntr, pmu_dev->cntr_assign_mask);
+}
+
+static inline void xgene_pmu_mask_int(struct xgene_pmu *xgene_pmu)
+{
+	writel(PCPPMU_INTENMASK, xgene_pmu->pcppmu_csr + PCPPMU_INTMASK_REG);
+}
+
+static inline void xgene_pmu_unmask_int(struct xgene_pmu *xgene_pmu)
+{
+	writel(PCPPMU_INTCLRMASK, xgene_pmu->pcppmu_csr + PCPPMU_INTMASK_REG);
+}
+
+static inline u32 xgene_pmu_read_counter(struct xgene_pmu_dev *pmu_dev, int idx)
+{
+	return readl(pmu_dev->inf->csr + PMU_PMEVCNTR0 + (4 * idx));
+}
+
+static inline void
+xgene_pmu_write_counter(struct xgene_pmu_dev *pmu_dev, int idx, u32 val)
+{
+	writel(val, pmu_dev->inf->csr + PMU_PMEVCNTR0 + (4 * idx));
+}
+
+static inline void
+xgene_pmu_write_evttype(struct xgene_pmu_dev *pmu_dev, int idx, u32 val)
+{
+	writel(val, pmu_dev->inf->csr + PMU_PMEVTYPER0 + (4 * idx));
+}
+
+static inline void
+xgene_pmu_write_agentmsk(struct xgene_pmu_dev *pmu_dev, u32 val)
+{
+	writel(val, pmu_dev->inf->csr + PMU_PMAMR0);
+}
+
+static inline void
+xgene_pmu_write_agent1msk(struct xgene_pmu_dev *pmu_dev, u32 val)
+{
+	writel(val, pmu_dev->inf->csr + PMU_PMAMR1);
+}
+
+static inline void
+xgene_pmu_enable_counter(struct xgene_pmu_dev *pmu_dev, int idx)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMCNTENSET);
+	val |= 1 << idx;
+	writel(val, pmu_dev->inf->csr + PMU_PMCNTENSET);
+}
+
+static inline void
+xgene_pmu_disable_counter(struct xgene_pmu_dev *pmu_dev, int idx)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMCNTENCLR);
+	val |= 1 << idx;
+	writel(val, pmu_dev->inf->csr + PMU_PMCNTENCLR);
+}
+
+static inline void
+xgene_pmu_enable_counter_int(struct xgene_pmu_dev *pmu_dev, int idx)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMINTENSET);
+	val |= 1 << idx;
+	writel(val, pmu_dev->inf->csr + PMU_PMINTENSET);
+}
+
+static inline void
+xgene_pmu_disable_counter_int(struct xgene_pmu_dev *pmu_dev, int idx)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMINTENCLR);
+	val |= 1 << idx;
+	writel(val, pmu_dev->inf->csr + PMU_PMINTENCLR);
+}
+
+static inline void xgene_pmu_reset_counters(struct xgene_pmu_dev *pmu_dev)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMCR);
+	val |= PMU_PMCR_P;
+	writel(val, pmu_dev->inf->csr + PMU_PMCR);
+}
+
+static inline void xgene_pmu_start_counters(struct xgene_pmu_dev *pmu_dev)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMCR);
+	val |= PMU_PMCR_E;
+	writel(val, pmu_dev->inf->csr + PMU_PMCR);
+}
+
+static inline void xgene_pmu_stop_counters(struct xgene_pmu_dev *pmu_dev)
+{
+	u32 val;
+
+	val = readl(pmu_dev->inf->csr + PMU_PMCR);
+	val &= ~PMU_PMCR_E;
+	writel(val, pmu_dev->inf->csr + PMU_PMCR);
+}
+
+static void xgene_perf_pmu_enable(struct pmu *pmu)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(pmu);
+	int enabled = bitmap_weight(pmu_dev->cntr_assign_mask,
+			pmu_dev->max_counters);
+
+	if (!enabled)
+		return;
+
+	xgene_pmu_start_counters(pmu_dev);
+}
+
+static void xgene_perf_pmu_disable(struct pmu *pmu)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(pmu);
+
+	xgene_pmu_stop_counters(pmu_dev);
+}
+
+static int xgene_perf_event_init(struct perf_event *event)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+	struct perf_event *sibling;
+
+	/* Test the event attr type check for PMU enumeration */
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	/*
+	 * SOC PMU counters are shared across all cores.
+	 * Therefore, it does not support per-process mode.
+	 * Also, it does not support event sampling mode.
+	 */
+	if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
+		return -EINVAL;
+
+	/* SOC counters do not have usr/os/guest/host bits */
+	if (event->attr.exclude_user || event->attr.exclude_kernel ||
+	    event->attr.exclude_host || event->attr.exclude_guest)
+		return -EINVAL;
+
+	if (event->cpu < 0)
+		return -EINVAL;
+	/*
+	 * Many perf core operations (eg. events rotation) operate on a
+	 * single CPU context. This is obvious for CPU PMUs, where one
+	 * expects the same sets of events being observed on all CPUs,
+	 * but can lead to issues for off-core PMUs, where each
+	 * event could be theoretically assigned to a different CPU. To
+	 * mitigate this, we enforce CPU assignment to one, selected
+	 * processor (the one described in the "cpumask" attribute).
+	 */
+	event->cpu = cpumask_first(&pmu_dev->parent->cpu);
+
+	hw->config = event->attr.config;
+	/*
+	 * Each bit of the config1 field represents an agent from which the
+	 * request of the event come. The event is counted only if it's caused
+	 * by a request of an agent has the bit cleared.
+	 * By default, the event is counted for all agents.
+	 */
+	hw->config_base = event->attr.config1;
+
+	/*
+	 * We must NOT create groups containing mixed PMUs, although software
+	 * events are acceptable
+	 */
+	if (event->group_leader->pmu != event->pmu &&
+			!is_software_event(event->group_leader))
+		return -EINVAL;
+
+	list_for_each_entry(sibling, &event->group_leader->sibling_list,
+			group_entry)
+		if (sibling->pmu != event->pmu &&
+				!is_software_event(sibling))
+			return -EINVAL;
+
+	return 0;
+}
+
+static void xgene_perf_enable_event(struct perf_event *event)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+
+	xgene_pmu_write_evttype(pmu_dev, GET_CNTR(event), GET_EVENTID(event));
+	xgene_pmu_write_agentmsk(pmu_dev, ~((u32)GET_AGENTID(event)));
+	if (pmu_dev->inf->type == PMU_TYPE_IOB)
+		xgene_pmu_write_agent1msk(pmu_dev, ~((u32)GET_AGENT1ID(event)));
+
+	xgene_pmu_enable_counter(pmu_dev, GET_CNTR(event));
+	xgene_pmu_enable_counter_int(pmu_dev, GET_CNTR(event));
+}
+
+static void xgene_perf_disable_event(struct perf_event *event)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+
+	xgene_pmu_disable_counter(pmu_dev, GET_CNTR(event));
+	xgene_pmu_disable_counter_int(pmu_dev, GET_CNTR(event));
+}
+
+static void xgene_perf_event_set_period(struct perf_event *event)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+	/*
+	 * The X-Gene PMU counters have a period of 2^32. To account for the
+	 * possiblity of extreme interrupt latency we program for a period of
+	 * half that. Hopefully we can handle the interrupt before another 2^31
+	 * events occur and the counter overtakes its previous value.
+	 */
+	u64 val = 1ULL << 31;
+
+	local64_set(&hw->prev_count, val);
+	xgene_pmu_write_counter(pmu_dev, hw->idx, (u32) val);
+}
+
+static void xgene_perf_event_update(struct perf_event *event)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+	u64 delta, prev_raw_count, new_raw_count;
+
+again:
+	prev_raw_count = local64_read(&hw->prev_count);
+	new_raw_count = xgene_pmu_read_counter(pmu_dev, GET_CNTR(event));
+
+	if (local64_cmpxchg(&hw->prev_count, prev_raw_count,
+			    new_raw_count) != prev_raw_count)
+		goto again;
+
+	delta = (new_raw_count - prev_raw_count) & pmu_dev->max_period;
+
+	local64_add(delta, &event->count);
+}
+
+static void xgene_perf_read(struct perf_event *event)
+{
+	xgene_perf_event_update(event);
+}
+
+static void xgene_perf_start(struct perf_event *event, int flags)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+
+	if (WARN_ON_ONCE(!(hw->state & PERF_HES_STOPPED)))
+		return;
+
+	WARN_ON_ONCE(!(hw->state & PERF_HES_UPTODATE));
+	hw->state = 0;
+
+	xgene_perf_event_set_period(event);
+
+	if (flags & PERF_EF_RELOAD) {
+		u64 prev_raw_count =  local64_read(&hw->prev_count);
+
+		xgene_pmu_write_counter(pmu_dev, GET_CNTR(event),
+					(u32) prev_raw_count);
+	}
+
+	xgene_perf_enable_event(event);
+	perf_event_update_userpage(event);
+}
+
+static void xgene_perf_stop(struct perf_event *event, int flags)
+{
+	struct hw_perf_event *hw = &event->hw;
+	u64 config;
+
+	if (hw->state & PERF_HES_UPTODATE)
+		return;
+
+	xgene_perf_disable_event(event);
+	WARN_ON_ONCE(hw->state & PERF_HES_STOPPED);
+	hw->state |= PERF_HES_STOPPED;
+
+	if (hw->state & PERF_HES_UPTODATE)
+		return;
+
+	config = hw->config;
+	xgene_perf_read(event);
+	hw->state |= PERF_HES_UPTODATE;
+}
+
+static int xgene_perf_add(struct perf_event *event, int flags)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+
+	hw->state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
+
+	/* Allocate an event counter */
+	hw->idx = get_next_avail_cntr(pmu_dev);
+	if (hw->idx < 0)
+		return -EAGAIN;
+
+	/* Update counter event pointer for Interrupt handler */
+	pmu_dev->pmu_counter_event[hw->idx] = event;
+
+	if (flags & PERF_EF_START)
+		xgene_perf_start(event, PERF_EF_RELOAD);
+
+	return 0;
+}
+
+static void xgene_perf_del(struct perf_event *event, int flags)
+{
+	struct xgene_pmu_dev *pmu_dev = to_pmu_dev(event->pmu);
+	struct hw_perf_event *hw = &event->hw;
+
+	xgene_perf_stop(event, PERF_EF_UPDATE);
+
+	/* clear the assigned counter */
+	clear_avail_cntr(pmu_dev, GET_CNTR(event));
+
+	perf_event_update_userpage(event);
+	pmu_dev->pmu_counter_event[hw->idx] = NULL;
+}
+
+static int xgene_init_perf(struct xgene_pmu_dev *pmu_dev, char *name)
+{
+	struct xgene_pmu *xgene_pmu;
+
+	pmu_dev->max_period = PMU_CNT_MAX_PERIOD - 1;
+	/* First version PMU supports only single event counter */
+	xgene_pmu = pmu_dev->parent;
+	if (xgene_pmu->version == PCP_PMU_V1)
+		pmu_dev->max_counters = 1;
+	else
+		pmu_dev->max_counters = PMU_MAX_COUNTERS;
+
+	/* Perf driver registration */
+	pmu_dev->pmu = (struct pmu) {
+		.attr_groups	= pmu_dev->attr_groups,
+		.task_ctx_nr	= perf_invalid_context,
+		.pmu_enable	= xgene_perf_pmu_enable,
+		.pmu_disable	= xgene_perf_pmu_disable,
+		.event_init	= xgene_perf_event_init,
+		.add		= xgene_perf_add,
+		.del		= xgene_perf_del,
+		.start		= xgene_perf_start,
+		.stop		= xgene_perf_stop,
+		.read		= xgene_perf_read,
+	};
+
+	/* Hardware counter init */
+	xgene_pmu_stop_counters(pmu_dev);
+	xgene_pmu_reset_counters(pmu_dev);
+
+	return perf_pmu_register(&pmu_dev->pmu, name, -1);
+}
+
+static int
+xgene_pmu_dev_add(struct xgene_pmu *xgene_pmu, struct xgene_pmu_dev_ctx *ctx)
+{
+	struct device *dev = xgene_pmu->dev;
+	struct xgene_pmu_dev *pmu;
+	int rc;
+
+	pmu = devm_kzalloc(dev, sizeof(*pmu), GFP_KERNEL);
+	if (!pmu)
+		return -ENOMEM;
+	pmu->parent = xgene_pmu;
+	pmu->inf = &ctx->inf;
+	ctx->pmu_dev = pmu;
+
+	switch (pmu->inf->type) {
+	case PMU_TYPE_L3C:
+		pmu->attr_groups = l3c_pmu_attr_groups;
+		break;
+	case PMU_TYPE_IOB:
+		pmu->attr_groups = iob_pmu_attr_groups;
+		break;
+	case PMU_TYPE_MCB:
+		if (!(xgene_pmu->mcb_active_mask & pmu->inf->enable_mask))
+			goto dev_err;
+		pmu->attr_groups = mcb_pmu_attr_groups;
+		break;
+	case PMU_TYPE_MC:
+		if (!(xgene_pmu->mc_active_mask & pmu->inf->enable_mask))
+			goto dev_err;
+		pmu->attr_groups = mc_pmu_attr_groups;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	rc = xgene_init_perf(pmu, ctx->name);
+	if (rc) {
+		dev_err(dev, "%s PMU: Failed to init perf driver\n", ctx->name);
+		goto dev_err;
+	}
+
+	dev_info(dev, "%s PMU registered\n", ctx->name);
+
+	return rc;
+
+dev_err:
+	devm_kfree(dev, pmu);
+	return -ENODEV;
+}
+
+static void _xgene_pmu_isr(int irq, struct xgene_pmu_dev *pmu_dev)
+{
+	struct xgene_pmu *xgene_pmu = pmu_dev->parent;
+	u32 pmovsr;
+	int idx;
+
+	pmovsr = readl(pmu_dev->inf->csr + PMU_PMOVSR) & PMU_OVERFLOW_MASK;
+	if (!pmovsr)
+		return;
+
+	/* Clear interrupt flag */
+	if (xgene_pmu->version == PCP_PMU_V1)
+		writel(0x0, pmu_dev->inf->csr + PMU_PMOVSR);
+	else
+		writel(pmovsr, pmu_dev->inf->csr + PMU_PMOVSR);
+
+	for (idx = 0; idx < PMU_MAX_COUNTERS; idx++) {
+		struct perf_event *event = pmu_dev->pmu_counter_event[idx];
+		int overflowed = pmovsr & BIT(idx);
+
+		/* Ignore if we don't have an event. */
+		if (!event || !overflowed)
+			continue;
+		xgene_perf_event_update(event);
+		xgene_perf_event_set_period(event);