Skip to contentSkip to conversation

IMC Instrumentation Support

PATCH New Hemant Kumar 1 participant 11 messages Feb 20, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v4 cover IMC Instrumentation Support

Power 9 has In-Memory-Collection (IMC) infrastructure which contains various Performance Monitoring Units (PMUs) at Nest level (these are on-chip but off-core), Core level and Thread level.

The Nest PMU counters are handled by a Nest IMC microcode which runs in the OCC (On-Chip Controller) complex. The microcode collects the counter data and moves the nest IMC counter data to memory.

The Core and Thread IMC PMU counters are handled in the core. Core level PMU counters give us the IMC counters' data per core and thread level PMU counters give us the IMC counters' data per CPU thread.

This patchset enables the nest IMC, core IMC and thread IMC PMUs and is based on the initial work done by Madhavan Srinivasan. "Nest Instrumentation Support" :
https://lists.ozlabs.org/pipermail/linuxppc-dev/2015-August/132078.html

v1 for this patchset can be found here :
https://lwn.net/Articles/705475/

Nest events:
Per-chip nest instrumentation provides various per-chip metrics such as memory, powerbus, Xlink and Alink bandwidth.

Core events:
Per-core IMC instrumentation provides various per-core metrics such as non-idle cycles, non-idle instructions, various cache and memory related metrics etc.

Thread events:
All the events for thread level are same as core level with the difference being in the domain. These are per-cpu metrics.

PMU Events' Information:
OPAL obtains the IMC PMU and event information from the IMC Catalog and passes on to the kernel via the device tree. The events' information contains :
- Event name
- Event Offset
- Event description
and, maybe :
- Event scale
- Event unit

Some PMUs may have a common scale and unit values for all their supported events. For those cases, the scale and unit properties for those events must be inherited from the PMU.

The event offset in the memory is where the counter data gets accumulated.

The OPAL-side patches are posted upstream :
https://lists.ozlabs.org/pipermail/skiboot/2017-January/005979.html

The kernel discovers the IMC counters information in the device tree at the "imc-counters" device node which has a compatible field "ibm,opal-in-memory-counters".

Parsing of the Events' information:
To parse the IMC PMUs and events information, the kernel has to discover the "imc-counters" node and walk through the pmu and event nodes.

Here is an excerpt of the dt showing the imc-counters with mcs0 (nest), core and thread node:
/dts-v1/;

[...]

/dts-v1/;

/ {
        name = "";
        compatible = "ibm,opal-in-memory-counters";
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        imc-nest-offset = <0x320000>;
        imc-nest-size = <0x30000>;
        version-id = "";
 
        NEST_MCS: nest-mcs-events {
                #address-cells = <0x1>;
                #size-cells = <0x1>;
 
                event@0 {
                        event-name = "RRTO_QFULL_NO_DISP" ;
                        reg = <0x0 0x8>;
                        desc = "RRTO not dispatched in MCS0 due to capacity - pulses once for each time a valid RRTO op is not dispatched due to a command list full condition" ;
                };
                event@8 {
                        event-name = "WRTO_QFULL_NO_DISP" ;
                        reg = <0x8 0x8>;
                        desc = "WRTO not dispatched in MCS0 due to capacity - pulses once for each time a valid WRTO op is not dispatched due to a command list full condition" ;
                };
		[...]
	mcs0 {
                compatible = "ibm,imc-counters-nest";
                events-prefix = "PM_MCS0_";
                unit = "";
                scale = "";
                reg = <0x118 0x8>;
                events = < &NEST_MCS >;
        };
 
        mcs1 {
                compatible = "ibm,imc-counters-nest";
                events-prefix = "PM_MCS1_";
                unit = "";
                scale = "";
                reg = <0x198 0x8>;
                events = < &NEST_MCS >;
        };
	[...]

	CORE_EVENTS: core-events {
                #address-cells = <0x1>;
                #size-cells = <0x1>;
 
                event@e0 {
                        event-name = "0THRD_NON_IDLE_PCYC" ;
                        reg = <0xe0 0x8>;
                        desc = "The number of processor cycles when all threads are idle" ;
                };
                event@120 {
                        event-name = "1THRD_NON_IDLE_PCYC" ;
                        reg = <0x120 0x8>;
                        desc = "The number of processor cycles when exactly one SMT thread is executing non-idle code" ;
                };
		[...]
       core {
                compatible = "ibm,imc-counters-core";
                events-prefix = "CPM_";
                unit = "";
                scale = "";
                reg = <0x0 0x8>;
                events = < &CORE_EVENTS >;
        };
 
        thread {
                compatible = "ibm,imc-counters-core";
                events-prefix = "CPM_";
                unit = "";
                scale = "";
                reg = <0x0 0x8>;
                events = < &CORE_EVENTS >;
        };
}; 
Quoted 1 line

From the device tree, the kernel parses the PMUs and their events'

information.

After parsing the IMC PMUs and their events, the PMUs and their attributes are registered in the kernel.

This patchset (patches 9 and 10) configure the thread level IMC PMUs to count for tasks, which give us the thread level metric values per task.

Example Usage :

 # perf list

[...]
nest_mcs0/PM_MCS_DOWN_128B_DATA_XFER_MC0/ [Kernel PMU event]
nest_mcs0/PM_MCS_DOWN_128B_DATA_XFER_MC0_LAST_SAMPLE/ [Kernel PMU event] [...]
core_imc/CPM_NON_IDLE_INST/ [Kernel PMU event]
core_imc/CPM_NON_IDLE_PCYC/ [Kernel PMU event]
[...]
thread_imc/CPM_NON_IDLE_INST/ [Kernel PMU event]
thread_imc/CPM_NON_IDLE_PCYC/ [Kernel PMU event]

To see per chip data for nest_mcs0/PM_MCS_DOWN_128B_DATA_XFER_MC0/ :

 # perf stat -e "nest_mcs0/PM_MCS_DOWN_128B_DATA_XFER_MC0/" -a --per-socket

To see non-idle instructions for core 0 :

 # ./perf stat -e "core_imc/CPM_NON_IDLE_INST/" -C 0 -I 1000

To see non-idle instructions for a "make" :

 # ./perf stat -e "thread_imc/CPM_NON_IDLE_PCYC/" make

Comments/feedback/suggestions are welcome.

Changelog:
v3 -> v4 :
- Changed the events parser code to discover the PMU and events because of the changed format of the IMC DTS file (Patch 3).
- Implemented the two TODOs to include core and thread IMC support with this patchset (Patches 7 through 10).
- Changed the CPU hotplug code of Nest IMC PMUs to include a new state CPUHP_AP_PERF_POWERPC_NEST_ONLINE (Patch 6).
v2 -> v3 :
- Changed all references for IMA (In-Memory Accumulation) to IMC (In-Memory Collection).
v1 -> v2 :
- Account for the cases where a PMU can have a common scale and unit values for all its supported events (Patch 3/6).
- Fixed a Build error (for maple_defconfig) by enabling imc_pmu.o only for CONFIG_PPC_POWERNV=y (Patch 4/6)
- Read from the "event-name" property instead of "name" for an event node (Patch 3/6).

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>

Hemant Kumar (10):
powerpc/powernv: Data structure and macros definitions powerpc/powernv: Autoload IMC device driver module powerpc/powernv: Detect supported IMC units and its events powerpc/perf: Add event attribute and group to IMC pmus powerpc/perf: Generic imc pmu event functions powerpc/perf: IMC pmu cpumask and cpu hotplug support powerpc/powernv: Core IMC events detection powerpc/perf: PMU functions for Core IMC and hotplugging powerpc/powernv: Thread IMC events detection powerpc/perf: Thread IMC PMU functions

 arch/powerpc/include/asm/imc-pmu.h             |  83 +++
 arch/powerpc/include/asm/opal-api.h            |  11 +-
 arch/powerpc/include/asm/opal.h                |   5 +
 arch/powerpc/perf/Makefile                     |   6 +-
 arch/powerpc/perf/imc-pmu.c                    | 775 +++++++++++++++++++++++++
 arch/powerpc/platforms/powernv/Makefile        |   2 +-
 arch/powerpc/platforms/powernv/opal-imc.c      | 553 ++++++++++++++++++
 arch/powerpc/platforms/powernv/opal-wrappers.S |   2 +
 arch/powerpc/platforms/powernv/opal.c          |  13 +
 include/linux/cpuhotplug.h                     |   2 +
 10 files changed, 1449 insertions(+), 3 deletions(-)
 create mode 100644 arch/powerpc/include/asm/imc-pmu.h
 create mode 100644 arch/powerpc/perf/imc-pmu.c
 create mode 100644 arch/powerpc/platforms/powernv/opal-imc.c
Signature · 1 line
2.7.4
#1 lore raw

PATCH v4 1/10 powerpc/powernv: Data structure and macros definitions

Create new header file "imc-pmu.h" to add the data structures and macros needed for IMC pmu support.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h | 73 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100644 arch/powerpc/include/asm/imc-pmu.h
arch/powerpc/include/asm/imc-pmu.h new file +73 -0
new file mode 100644
index 0000000..3232322
@@ -0,0 +1,73 @@
+#ifndef PPC_POWERNV_IMC_PMU_DEF_H
+#define PPC_POWERNV_IMC_PMU_DEF_H
+
+/*
+ * IMC Nest Performance Monitor counter support.
+ *
+ * Copyright (C) 2016 Madhavan Srinivasan, IBM Corporation.
+ *           (C) 2016 Hemant K Shaw, IBM Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation.
+ */
+
+#include <linux/perf_event.h>
+#include <linux/slab.h>
+#include <linux/of.h>
+#include <linux/io.h>
+#include <asm/opal.h>
+
+#define IMC_MAX_CHIPS			32
+#define IMC_MAX_PMUS			32
+#define IMC_MAX_PMU_NAME_LEN		256
+
+#define NEST_IMC_ENGINE_START		1
+#define NEST_IMC_ENGINE_STOP		0
+#define NEST_MAX_PAGES			16
+
+#define NEST_IMC_PRODUCTION_MODE	1
+
+#define IMC_DTB_COMPAT		"ibm,opal-in-memory-counters"
+#define IMC_DTB_NEST_COMPAT	"ibm,imc-counters-nest"
+
+/*
+ * Structure to hold per chip specific memory address
+ * information for nest pmus. Nest Counter data are exported
+ * in per-chip reserved memory region by the PORE Engine.
+ */
+struct perchip_nest_info {
+	u32 chip_id;
+	u64 pbase;
+	u64 vbase[NEST_MAX_PAGES];
+	u64 size;
+};
+
+/*
+ * Place holder for nest pmu events and values.
+ */
+struct imc_events {
+	char *ev_name;
+	char *ev_value;
+};
+
+/*
+ * Device tree parser code detects IMC pmu support and
+ * registers new IMC pmus. This structure will
+ * hold the pmu functions and attrs for each imc pmu and
+ * will be referenced at the time of pmu registration.
+ */
+struct imc_pmu {
+	struct pmu pmu;
+	int domain;
+	const struct attribute_group *attr_groups[4];
+};
+
+/*
+ * Domains for IMC PMUs
+ */
+#define IMC_DOMAIN_NEST		1
+
+#define UNKNOWN_DOMAIN		-1
+
+#endif /* PPC_POWERNV_IMC_PMU_DEF_H */
Signature · 1 line
2.7.4
#2 lore raw

PATCH v4 2/10 powerpc/powernv: Autoload IMC device driver module

In reply to Hemant Kumar

This patch does three things :
- Enables "opal.c" to create a platform device for the IMC interface according to the appropriate compatibility string.
- Find the reserved-memory region details from the system device tree and get the base address of HOMER region address for each chip.
- We also get the Nest PMU counter data offsets (in the HOMER region) and their sizes. The offsets for the counters' data are fixed and won't change from chip to chip.

The device tree parsing logic is separated from the PMU creation functions (which is done in subsequent patches). Right now, only Nest units are taken care of.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/platforms/powernv/Makefile   |   2 +-
 arch/powerpc/platforms/powernv/opal-imc.c | 117 ++++++++++++++++++++++++++++++
 arch/powerpc/platforms/powernv/opal.c     |  13 ++++
 3 files changed, 131 insertions(+), 1 deletion(-)
 create mode 100644 arch/powerpc/platforms/powernv/opal-imc.c
arch/powerpc/platforms/powernv/Makefile +1 -1
index b5d98cb..44909fe 100644
@@ -2,7 +2,7 @@ obj-y			+= setup.o opal-wrappers.o opal.o opal-async.o idle.o
 obj-y			+= opal-rtc.o opal-nvram.o opal-lpc.o opal-flash.o
 obj-y			+= rng.o opal-elog.o opal-dump.o opal-sysparam.o opal-sensor.o
 obj-y			+= opal-msglog.o opal-hmi.o opal-power.o opal-irqchip.o
-obj-y			+= opal-kmsg.o
+obj-y			+= opal-kmsg.o opal-imc.o
 
 obj-$(CONFIG_SMP)	+= smp.o subcore.o subcore-asm.o
 obj-$(CONFIG_PCI)	+= pci.o pci-ioda.o npu-dma.o
arch/powerpc/platforms/powernv/opal-imc.c new file +117 -0

117 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal.c +13 -0
index 86d9fde..b2952f8 100644
@@ -30,6 +30,7 @@
 #include <asm/opal.h>
 #include <asm/firmware.h>
 #include <asm/mce.h>
+#include <asm/imc-pmu.h>
 
 #include "powernv.h"
 
@@ -640,6 +641,15 @@ static void opal_pdev_init(const char *compatible)
 		of_platform_device_create(np, NULL, NULL);
 }
 
+static void opal_imc_init_dev(void)
+{
+	struct device_node *np;
+
+	np = of_find_compatible_node(NULL, NULL, IMC_DTB_COMPAT);
+	if (np)
+		of_platform_device_create(np, NULL, NULL);
+}
+
 static int kopald(void *unused)
 {
 	unsigned long timeout = msecs_to_jiffies(opal_heartbeat) + 1;
@@ -713,6 +723,9 @@ static int __init opal_init(void)
 	/* Setup a heatbeat thread if requested by OPAL */
 	opal_init_heartbeat();
 
+	/* Detect IMC pmu counters support and create PMUs */
+	opal_imc_init_dev();
+
 	/* Create leds platform devices */
 	leds = of_find_node_by_path("/ibm,opal/leds");
 	if (leds) {
Signature · 1 line
2.7.4
#3 lore raw

PATCH v4 3/10 powerpc/powernv: Detect supported IMC units and its events

In reply to Hemant Kumar

Parse device tree to detect IMC units. Traverse through each IMC unit node to find supported events and corresponding unit/scale files (if any).

The device tree for IMC counters starts at the node :
"imc-counters". This node contains all the IMC PMU nodes and event nodes for these IMC PMUs. The PMU nodes have an "events" property which has a phandle value for the actual events node. The events are separated from the PMU nodes to abstract out the common events. For example, PMU node "mcs0", "mcs1" etc. will contain a pointer to "nest-mcs-events" since, the events are common between these PMUs. These events have a different prefix based on their relation to different PMUs, and hence, the PMU nodes themselves contain an "events-prefix" property. The value for this property concatenated to the event name, forms the actual event name. Also, the PMU have a "reg" field as the base offset for the events which belong to this PMU. This "reg" field is added to an event in the "events" node, which gives us the location of the counter data. Kernel code uses this offset as event configuration value.

Device tree parser code also looks for scale/unit property in the event node and passes on the value as an event attr for perf interface to use in the post processing by the perf tool. Some PMUs may have common scale and unit properties which implies that all events supported by this PMU inherit the scale and unit properties of the PMU itself. For those events, we need to set the common unit and scale values.

For failure to initialize any unit or any event, disable that unit and continue setting up the rest of them.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
---
 arch/powerpc/platforms/powernv/opal-imc.c | 385 ++++++++++++++++++++++++++++++
 1 file changed, 385 insertions(+)
arch/powerpc/platforms/powernv/opal-imc.c +385 -0
index ee2ae45..c58b893 100644
@@ -32,6 +32,390 @@
 #include <asm/imc-pmu.h>
 
 struct perchip_nest_info nest_perchip_info[IMC_MAX_CHIPS];
+struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
+
+static int imc_event_info(char *name, struct imc_events *events)
+{
+	char *buf;
+
+	/* memory for content */
+	buf = kzalloc(IMC_MAX_PMU_NAME_LEN, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	events->ev_name = name;
+	events->ev_value = buf;
+	return 0;
+}
+
+static int imc_event_info_str(struct property *pp, char *name,
+			       struct imc_events *events)
+{
+	int ret;
+
+	ret = imc_event_info(name, events);
+	if (ret)
+		return ret;
+
+	if (!pp->value || (strnlen(pp->value, pp->length) == pp->length) ||
+	   (pp->length > IMC_MAX_PMU_NAME_LEN))
+		return -EINVAL;
+	strncpy(events->ev_value, (const char *)pp->value, pp->length);
+
+	return 0;
+}
+
+static int imc_event_info_val(char *name, u32 val,
+			      struct imc_events *events)
+{
+	int ret;
+
+	ret = imc_event_info(name, events);
+	if (ret)
+		return ret;
+	sprintf(events->ev_value, "event=0x%x", val);
+
+	return 0;
+}
+
+static int set_event_property(struct property *pp, char *event_prop,
+			      struct imc_events *events, char *ev_name)
+{
+	char *buf;
+	int ret;
+
+	buf = kzalloc(IMC_MAX_PMU_NAME_LEN, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	sprintf(buf, "%s.%s", ev_name, event_prop);
+	ret = imc_event_info_str(pp, buf, events);
+	if (ret) {
+		kfree(events->ev_name);
+		kfree(events->ev_value);
+	}
+
+	return ret;
+}
+
+/*
+ * imc_events_node_parser: Parse the event node "dev" and assign the parsed
+ *                         information to event "events".
+ *
+ * Parses the "reg" property of this event. "reg" gives us the event offset.
+ * Also, parse the "scale" and "unit" properties, if any.
+ */
+static int imc_events_node_parser(struct device_node *dev,
+				  struct imc_events *events,
+				  struct property *event_scale,
+				  struct property *event_unit,
+				  struct property *name_prefix,
+				  u32 reg)
+{
+	struct property *name, *pp;
+	char *ev_name;
+	u32 val;
+	int idx = 0, ret;
+
+	if (!dev)
+		return -EINVAL;
+
+	/*
+	 * Loop through each property of an event node
+	 */
+	name = of_find_property(dev, "event-name", NULL);
+	if (!name)
+		return -ENODEV;
+
+	if (!name->value ||
+	  (strnlen(name->value, name->length) == name->length) ||
+	  (name->length > IMC_MAX_PMU_NAME_LEN))
+		return -EINVAL;
+
+	ev_name = kzalloc(IMC_MAX_PMU_NAME_LEN, GFP_KERNEL);
+	if (!ev_name)
+		return -ENOMEM;
+
+	snprintf(ev_name, IMC_MAX_PMU_NAME_LEN, "%s%s",
+		 (char *)name_prefix->value,
+		 (char *)name->value);
+
+	/*
+	 * Parse each property of this event node "dev". Property "reg" has
+	 * the offset which is assigned to the event name. Other properties
+	 * like "scale" and "unit" are assigned to event.scale and event.unit
+	 * accordingly.
+	 */
+	for_each_property_of_node(dev, pp) {
+		/*
+		 * If there is an issue in parsing a single property of
+		 * this event, we just clean up the buffers, but we still
+		 * continue to parse.
+		 */
+		if (strncmp(pp->name, "reg", 3) == 0) {
+			of_property_read_u32(dev, pp->name, &val);
+			val += reg;
+			ret = imc_event_info_val(ev_name, val, &events[idx]);
+			if (ret) {
+				kfree(events[idx].ev_name);
+				kfree(events[idx].ev_value);
+				continue;
+			}
+			/*
+			 * If the common scale and unit properties available,
+			 * then, assign them to this event
+			 */
+			if (event_scale) {
+				idx++;
+				ret = set_event_property(event_scale, "scale",
+							 &events[idx],
+							 ev_name);
+				if (ret)
+					continue;
+				idx++;
+			}
+			if (event_unit) {
+				ret = set_event_property(event_unit, "unit",
+							 &events[idx],
+							 ev_name);
+				if (ret)
+					continue;
+			}
+			idx++;
+		} else if (strncmp(pp->name, "unit", 4) == 0) {
+			ret = set_event_property(pp, "unit", &events[idx],
+						 ev_name);
+			if (ret)
+				continue;
+			idx++;
+		} else if (strncmp(pp->name, "scale", 5) == 0) {
+			ret = set_event_property(pp, "scale", &events[idx],
+						 ev_name);
+			if (ret)
+				continue;
+			idx++;
+		}
+	}
+
+	return idx;
+}
+
+/*
+ * imc_get_domain : Returns the domain for pmu "pmu_dev".
+ */
+int imc_get_domain(struct device_node *pmu_dev)
+{
+	if (of_device_is_compatible(pmu_dev, IMC_DTB_NEST_COMPAT))
+		return IMC_DOMAIN_NEST;
+	else
+		return UNKNOWN_DOMAIN;
+}
+
+/*
+ * get_nr_children : Returns the number of children for a pmu device node.
+ */
+static int get_nr_children(struct device_node *pmu_node)
+{
+	struct device_node *child;
+	int i = 0;
+
+	for_each_child_of_node(pmu_node, child)
+		i++;
+	return i;
+}
+
+/*
+ * imc_free_events : Cleanup the "events" list having "nr_entries" entries.
+ */
+static void imc_free_events(struct imc_events *events, int nr_entries)
+{
+	int i;
+
+	/* Nothing to clean, return */
+	if (!events)
+		return;
+	for (i = 0; i < nr_entries; i++) {
+		kfree(events[i].ev_name);
+		kfree(events[i].ev_value);
+	}
+
+	kfree(events);
+}
+
+/*
+ * imc_pmu_create : Takes the parent device which is the pmu unit and a
+ *                  pmu_index as the inputs.
+ * Allocates memory for the pmu, sets up its domain (NEST or CORE), and
+ * allocates memory for the events supported by this pmu. Assigns a name for
+ * the pmu. Calls imc_events_node_parser() to setup the individual events.
+ * If everything goes fine, it calls, init_imc_pmu() to setup the pmu device
+ * and register it.
+ */
+static int imc_pmu_create(struct device_node *parent, int pmu_index)
+{
+	struct device_node *ev_node = NULL, *dir = NULL;
+	struct imc_events *events;
+	struct imc_pmu *pmu_ptr;
+	u32 prop, reg;
+	struct property *pp, *scale_pp, *unit_pp, *name_prefix;
+	char *buf;
+	int idx = 0, ret = 0, nr_children = 0;
+
+	if (!parent)
+		return -EINVAL;
+
+	/* memory for pmu */
+	pmu_ptr = kzalloc(sizeof(struct imc_pmu), GFP_KERNEL);
+	if (!pmu_ptr)
+		return -ENOMEM;
+
+	pmu_ptr->domain = imc_get_domain(parent);
+	if (pmu_ptr->domain == UNKNOWN_DOMAIN)
+		goto free_pmu;
+
+	/* Needed for hotplug/migration */
+	per_nest_pmu_arr[pmu_index] = pmu_ptr;
+
+	/*
+	 * "events" property inside a PMU node contains the phandle value
+	 * for the actual events node. The "events" node for the IMC PMU
+	 * is not in this node, rather inside "imc-counters" node, since,
+	 * we want to factor out the common events (thereby, reducing the
+	 * size of the device tree)
+	 */
+	of_property_read_u32(parent, "events", &prop);
+	if (!prop)
+		return -EINVAL;
+
+	/*
+	 * Fetch the actual node where the events for this PMU exist.
+	 */
+	dir = of_find_node_by_phandle(prop);
+	if (!dir)
+		return -EINVAL;
+
+	/*
+	 * Get the maximum no. of events in this node.
+	 * Multiply by 3 to account for .scale and .unit properties
+	 * This number suggests the amount of memory needed to setup the
+	 * events for this pmu.
+	 */
+	nr_children = get_nr_children(dir) * 3;
+
+	/* memory for pmu events */
+	events = kzalloc((sizeof(struct imc_events) * nr_children),
+			 GFP_KERNEL);
+	if (!events) {
+		ret = -ENOMEM;
+		goto free_pmu;
+	}
+
+	pp = of_find_property(parent, "name", NULL);
+	if (!pp) {
+		ret = -ENODEV;
+		goto free_events;
+	}
+
+	if (!pp->value ||
+	  (strnlen(pp->value, pp->length) == pp->length) ||
+	    (pp->length > IMC_MAX_PMU_NAME_LEN)) {
+		ret = -EINVAL;
+		goto free_events;
+	}
+
+	buf = kzalloc(IMC_MAX_PMU_NAME_LEN, GFP_KERNEL);
+	if (!buf) {
+		ret = -ENOMEM;
+		goto free_events;
+	}
+
+	/* Save the name to register it later */
+	sprintf(buf, "nest_%s", (char *)pp->value);
+	pmu_ptr->pmu.name = (char *)buf;
+
+	/*
+	 * Check if there is a common "scale" and "unit" properties inside
+	 * the PMU node for all the events supported by this PMU.
+	 */
+	scale_pp = of_find_property(parent, "scale", NULL);
+	unit_pp = of_find_property(parent, "unit", NULL);
+
+	/*
+	 * Get the event-prefix property from the PMU node
+	 * which needs to be attached with the event names.
+	 */
+	name_prefix = of_find_property(parent, "events-prefix", NULL);
+	if (!name_prefix)
+		return -ENODEV;
+
+	/*
+	 * "reg" property gives out the base offset of the counters data
+	 * for this PMU.
+	 */
+	of_property_read_u32(parent, "reg", &reg);
+
+	if (!name_prefix->value ||
+	   (strnlen(name_prefix->value, name_prefix->length) == name_prefix->length) ||
+	   (name_prefix->length > IMC_MAX_PMU_NAME_LEN))
+		return -EINVAL;
+
+	/* Loop through event nodes */
+	for_each_child_of_node(dir, ev_node) {
+		ret = imc_events_node_parser(ev_node, &events[idx], scale_pp,
+					     unit_pp, name_prefix, reg);
+		if (ret < 0) {
+			/* Unable to parse this event */
+			if (ret == -ENOMEM)
+				goto free_events;
+			continue;
+		}
+
+		/*
+		 * imc_event_node_parser will return number of
+		 * event entries created for this. This could include
+		 * event scale and unit files also.
+		 */
+		idx += ret;
+	}
+
+	return 0;
+
+free_events:
+	imc_free_events(events, idx);
+free_pmu:
+	kfree(pmu_ptr);
+	return ret;
+}
+
+/*
+ * imc_pmu_setup : Setup the IMC PMUs (children of "parent").
+ */
+static void imc_pmu_setup(struct device_node *parent)
+{
+	struct device_node *child;
+	int pmu_count = 0, rc = 0;
+	struct property *pp;
+
+	if (!parent)
+		return;
+
+	/* Setup all the IMC pmus */
+	for_each_child_of_node(parent, child) {
+		for_each_property_of_node(child, pp) {
+			/*
+			 * If there is a node with a "compatible" field,
+			 * that's a PMU node or else, its an events node.
+			 */
+			if (strncmp(pp->name, "compatible", 10)) {
+				rc = imc_pmu_create(child, pmu_count);
+				if (rc)
+					return;
+				pmu_count++;
+				break;
+			}
+		}
+	}
+}
 
 static int opal_imc_counters_probe(struct platform_device *pdev)
 {
@@ -93,6 +477,7 @@ static int opal_imc_counters_probe(struct platform_device *pdev)
 		} while (i < (pcni->size / PAGE_SIZE));
 	}
 
+	imc_pmu_setup(imc_dev);
 	return 0;
 err:
 	return -ENODEV;
Signature · 1 line
2.7.4
#4 lore raw

PATCH v4 4/10 powerpc/perf: Add event attribute and group to IMC pmus

In reply to Hemant Kumar

Device tree IMC driver code parses the IMC units and their events. It passes the information to IMC pmu code which is placed in powerpc/perf as "imc-pmu.c".

This patch creates only event attributes and attribute groups for the IMC pmus.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/perf/Makefile                |  6 +-
 arch/powerpc/perf/imc-pmu.c               | 96 +++++++++++++++++++++++++++++++
 arch/powerpc/platforms/powernv/opal-imc.c | 12 +++-
 3 files changed, 111 insertions(+), 3 deletions(-)
 create mode 100644 arch/powerpc/perf/imc-pmu.c
arch/powerpc/perf/Makefile +5 -1
index 4d606b9..d0d1f04 100644
@@ -2,10 +2,14 @@ subdir-ccflags-$(CONFIG_PPC_WERROR) := -Werror
 
 obj-$(CONFIG_PERF_EVENTS)	+= callchain.o perf_regs.o
 
+imc-$(CONFIG_PPC_POWERNV)       += imc-pmu.o
+
 obj-$(CONFIG_PPC_PERF_CTRS)	+= core-book3s.o bhrb.o
 obj64-$(CONFIG_PPC_PERF_CTRS)	+= power4-pmu.o ppc970-pmu.o power5-pmu.o \
 				   power5+-pmu.o power6-pmu.o power7-pmu.o \
-				   isa207-common.o power8-pmu.o power9-pmu.o
+				   isa207-common.o power8-pmu.o power9-pmu.o \
+				   $(imc-y)
+
 obj32-$(CONFIG_PPC_PERF_CTRS)	+= mpc7450-pmu.o
 
 obj-$(CONFIG_FSL_EMB_PERF_EVENT) += core-fsl-emb.o
arch/powerpc/perf/imc-pmu.c new file +96 -0

96 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal-imc.c +10 -2
index c58b893..ed1e091 100644
@@ -31,8 +31,11 @@
 #include <asm/cputable.h>
 #include <asm/imc-pmu.h>
 
-struct perchip_nest_info nest_perchip_info[IMC_MAX_CHIPS];
-struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
+extern struct perchip_nest_info nest_perchip_info[IMC_MAX_CHIPS];
+extern struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
+
+extern int init_imc_pmu(struct imc_events *events,
+			int idx, struct imc_pmu *pmu_ptr);
 
 static int imc_event_info(char *name, struct imc_events *events)
 {
@@ -378,6 +381,11 @@ static int imc_pmu_create(struct device_node *parent, int pmu_index)
 		idx += ret;
 	}
 
+	ret = init_imc_pmu(events, idx, pmu_ptr);
+	if (ret) {
+		pr_err("IMC PMU %s Register failed\n", pmu_ptr->pmu.name);
+		goto free_events;
+	}
 	return 0;
 
 free_events:
Signature · 1 line
2.7.4
#5 lore raw

PATCH v4 5/10 powerpc/perf: Generic imc pmu event functions

In reply to Hemant Kumar

Since, the IMC counters' data are periodically fed to a memory location, the functions to read/update, start/stop, add/del can be generic and can be used by all IMC PMU units.

This patch adds a set of generic imc pmu related event functions to be used by each imc pmu unit. Add code to setup format attribute and to register imc pmus. Add a event_init function for nest_imc events.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h        |   1 +
 arch/powerpc/perf/imc-pmu.c               | 121 ++++++++++++++++++++++++++++++
 arch/powerpc/platforms/powernv/opal-imc.c |  30 +++++++-
 3 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/imc-pmu.h +1 -0
index 3232322..7b58721 100644
@@ -70,4 +70,5 @@ struct imc_pmu {
 
 #define UNKNOWN_DOMAIN		-1
 
+int imc_get_domain(struct device_node *pmu_dev);
 #endif /* PPC_POWERNV_IMC_PMU_DEF_H */
arch/powerpc/perf/imc-pmu.c +121 -0

133 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal-imc.c +26 -4
index ed1e091..a65aa2d 100644
@@ -36,6 +36,7 @@ extern struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
 
 extern int init_imc_pmu(struct imc_events *events,
 			int idx, struct imc_pmu *pmu_ptr);
+u64 nest_max_offset;
 
 static int imc_event_info(char *name, struct imc_events *events)
 {
@@ -68,8 +69,25 @@ static int imc_event_info_str(struct property *pp, char *name,
 	return 0;
 }
 
+/*
+ * Updates the maximum offset for an event in the pmu with domain
+ * "pmu_domain". Right now, only nest domain is supported.
+ */
+static void update_max_value(u32 value, int pmu_domain)
+{
+	switch (pmu_domain) {
+	case IMC_DOMAIN_NEST:
+		if (nest_max_offset < value)
+			nest_max_offset = value;
+		break;
+	default:
+		/* Unknown domain, return */
+		return;
+	}
+}
+
 static int imc_event_info_val(char *name, u32 val,
-			      struct imc_events *events)
+			      struct imc_events *events, int pmu_domain)
 {
 	int ret;
 
@@ -77,6 +95,7 @@ static int imc_event_info_val(char *name, u32 val,
 	if (ret)
 		return ret;
 	sprintf(events->ev_value, "event=0x%x", val);
+	update_max_value(val, pmu_domain);
 
 	return 0;
 }
@@ -113,7 +132,8 @@ static int imc_events_node_parser(struct device_node *dev,
 				  struct property *event_scale,
 				  struct property *event_unit,
 				  struct property *name_prefix,
-				  u32 reg)
+				  u32 reg,
+				  int pmu_domain)
 {
 	struct property *name, *pp;
 	char *ev_name;
@@ -158,7 +178,8 @@ static int imc_events_node_parser(struct device_node *dev,
 		if (strncmp(pp->name, "reg", 3) == 0) {
 			of_property_read_u32(dev, pp->name, &val);
 			val += reg;
-			ret = imc_event_info_val(ev_name, val, &events[idx]);
+			ret = imc_event_info_val(ev_name, val, &events[idx],
+				pmu_domain);
 			if (ret) {
 				kfree(events[idx].ev_name);
 				kfree(events[idx].ev_value);
@@ -365,7 +386,8 @@ static int imc_pmu_create(struct device_node *parent, int pmu_index)
 	/* Loop through event nodes */
 	for_each_child_of_node(dir, ev_node) {
 		ret = imc_events_node_parser(ev_node, &events[idx], scale_pp,
-					     unit_pp, name_prefix, reg);
+					     unit_pp, name_prefix, reg,
+					     pmu_ptr->domain);
 		if (ret < 0) {
 			/* Unable to parse this event */
 			if (ret == -ENOMEM)
Signature · 1 line
2.7.4
#6 lore raw

PATCH v4 6/10 powerpc/perf: IMC pmu cpumask and cpu hotplug support

In reply to Hemant Kumar

Adds cpumask attribute to be used by each IMC pmu. Only one cpu (any online CPU) from each chip for nest PMUs is designated to read counters.

On CPU hotplug, dying CPU is checked to see whether it is one of the designated cpus, if yes, next online cpu from the same chip (for nest units) is designated as new cpu to read counters. For this purpose, we introduce a new state : CPUHP_AP_PERF_POWERPC_NEST_ONLINE.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/opal-api.h            |   3 +-
 arch/powerpc/include/asm/opal.h                |   3 +
 arch/powerpc/perf/imc-pmu.c                    | 163 ++++++++++++++++++++++++-
 arch/powerpc/platforms/powernv/opal-wrappers.S |   1 +
 include/linux/cpuhotplug.h                     |   1 +
 5 files changed, 169 insertions(+), 2 deletions(-)
arch/powerpc/include/asm/opal-api.h +2 -1
index a0aa285..e15fb20 100644
@@ -168,7 +168,8 @@
 #define OPAL_INT_SET_MFRR			125
 #define OPAL_PCI_TCE_KILL			126
 #define OPAL_NMMU_SET_PTCR			127
-#define OPAL_LAST				127
+#define OPAL_NEST_IMC_COUNTERS_CONTROL		128
+#define OPAL_LAST				128
 
 /* Device tree flags */
 
arch/powerpc/include/asm/opal.h +3 -0
index 1ff03a6..d93d082 100644
@@ -227,6 +227,9 @@ int64_t opal_pci_tce_kill(uint64_t phb_id, uint32_t kill_type,
 			  uint64_t dma_addr, uint32_t npages);
 int64_t opal_nmmu_set_ptcr(uint64_t chip_id, uint64_t ptcr);
 
+int64_t opal_nest_imc_counters_control(uint64_t mode, uint64_t value1,
+				uint64_t value2, uint64_t value3);
+
 /* Internal functions */
 extern int early_init_dt_scan_opal(unsigned long node, const char *uname,
 				   int depth, void *data);
arch/powerpc/perf/imc-pmu.c +162 -1

193 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal-wrappers.S +1 -0
index 6693f75..9be9a9c 100644
@@ -301,3 +301,4 @@ OPAL_CALL(opal_int_eoi,				OPAL_INT_EOI);
 OPAL_CALL(opal_int_set_mfrr,			OPAL_INT_SET_MFRR);
 OPAL_CALL(opal_pci_tce_kill,			OPAL_PCI_TCE_KILL);
 OPAL_CALL(opal_nmmu_set_ptcr,			OPAL_NMMU_SET_PTCR);
+OPAL_CALL(opal_nest_imc_counters_control,	OPAL_NEST_IMC_COUNTERS_CONTROL);
include/linux/cpuhotplug.h +1 -0
index bb790c4..8fac5ad 100644
@@ -138,6 +138,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_CCN_ONLINE,
 	CPUHP_AP_PERF_ARM_L2X0_ONLINE,
 	CPUHP_AP_PERF_ARM_QCOM_L2_ONLINE,
+	CPUHP_AP_PERF_POWERPC_NEST_ONLINE,
 	CPUHP_AP_WORKQUEUE_ONLINE,
 	CPUHP_AP_RCUTREE_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
Signature · 1 line
2.7.4
#7 lore raw

PATCH v4 7/10 powerpc/powernv: Core IMC events detection

In reply to Hemant Kumar

This patch adds support for detection of core IMC events along with the Nest IMC events. It adds a new domain IMC_DOMAIN_CORE and its determined with the help of the compatibility string "ibm,imc-counters-core" based on the IMC device tree.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h        |  2 ++
 arch/powerpc/perf/imc-pmu.c               |  3 +++
 arch/powerpc/platforms/powernv/opal-imc.c | 18 ++++++++++++++++--
 3 files changed, 21 insertions(+), 2 deletions(-)
arch/powerpc/include/asm/imc-pmu.h +2 -0
index 7b58721..59de083 100644
@@ -30,6 +30,7 @@
 
 #define IMC_DTB_COMPAT		"ibm,opal-in-memory-counters"
 #define IMC_DTB_NEST_COMPAT	"ibm,imc-counters-nest"
+#define IMC_DTB_CORE_COMPAT	"ibm,imc-counters-core"
 
 /*
  * Structure to hold per chip specific memory address
@@ -67,6 +68,7 @@ struct imc_pmu {
  * Domains for IMC PMUs
  */
 #define IMC_DOMAIN_NEST		1
+#define IMC_DOMAIN_CORE		2
 
 #define UNKNOWN_DOMAIN		-1
 
arch/powerpc/perf/imc-pmu.c +3 -0
index e46ff6d..9a0e3bc 100644
@@ -18,8 +18,11 @@ struct perchip_nest_info nest_perchip_info[IMC_MAX_CHIPS];
 struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
 static cpumask_t nest_imc_cpumask;
 
+struct imc_pmu *core_imc_pmu;
+
 /* Needed for sanity check */
 extern u64 nest_max_offset;
+extern u64 core_max_offset;
 
 PMU_FORMAT_ATTR(event, "config:0-20");
 static struct attribute *imc_format_attrs[] = {
arch/powerpc/platforms/powernv/opal-imc.c +16 -2
index a65aa2d..67ce873 100644
@@ -33,10 +33,12 @@
 
 extern struct perchip_nest_info nest_perchip_info[IMC_MAX_CHIPS];
 extern struct imc_pmu *per_nest_pmu_arr[IMC_MAX_PMUS];
+extern struct imc_pmu *core_imc_pmu;
 
 extern int init_imc_pmu(struct imc_events *events,
 			int idx, struct imc_pmu *pmu_ptr);
 u64 nest_max_offset;
+u64 core_max_offset;
 
 static int imc_event_info(char *name, struct imc_events *events)
 {
@@ -80,6 +82,10 @@ static void update_max_value(u32 value, int pmu_domain)
 		if (nest_max_offset < value)
 			nest_max_offset = value;
 		break;
+	case IMC_DOMAIN_CORE:
+		if (core_max_offset < value)
+			core_max_offset = value;
+		break;
 	default:
 		/* Unknown domain, return */
 		return;
@@ -231,6 +237,8 @@ int imc_get_domain(struct device_node *pmu_dev)
 {
 	if (of_device_is_compatible(pmu_dev, IMC_DTB_NEST_COMPAT))
 		return IMC_DOMAIN_NEST;
+	if (of_device_is_compatible(pmu_dev, IMC_DTB_CORE_COMPAT))
+		return IMC_DOMAIN_CORE;
 	else
 		return UNKNOWN_DOMAIN;
 }
@@ -298,7 +306,10 @@ static int imc_pmu_create(struct device_node *parent, int pmu_index)
 		goto free_pmu;
 
 	/* Needed for hotplug/migration */
-	per_nest_pmu_arr[pmu_index] = pmu_ptr;
+	if (pmu_ptr->domain == IMC_DOMAIN_CORE)
+		core_imc_pmu = pmu_ptr;
+	else if (pmu_ptr->domain == IMC_DOMAIN_NEST)
+		per_nest_pmu_arr[pmu_index] = pmu_ptr;
 
 	/*
 	 * "events" property inside a PMU node contains the phandle value
@@ -354,7 +365,10 @@ static int imc_pmu_create(struct device_node *parent, int pmu_index)
 	}
 
 	/* Save the name to register it later */
-	sprintf(buf, "nest_%s", (char *)pp->value);
+	if (pmu_ptr->domain == IMC_DOMAIN_NEST)
+		sprintf(buf, "nest_%s", (char *)pp->value);
+	else
+		sprintf(buf, "%s_imc", (char *)pp->value);
 	pmu_ptr->pmu.name = (char *)buf;
 
 	/*
Signature · 1 line
2.7.4
#8 lore raw

PATCH v4 8/10 powerpc/perf: PMU functions for Core IMC and hotplugging

In reply to Hemant Kumar

This patch adds the PMU function to initialize a core IMC event. It also adds cpumask initialization function for core IMC PMU. For initialization, a page of memory is allocated per core where the data for core IMC counters will be accumulated. The base address for this page is sent to OPAL via an OPAL call which initializes various SCOMs related to Core IMC initialization. Upon any errors, the pages are free'ed and core IMC counters are disabled using the same OPAL call.

For CPU hotplugging, a cpumask is initialized which contains an online CPU from each core. If a cpu goes offline, we check whether that cpu belongs to the core imc cpumask, if yes, then, we migrate the PMU context to any other online cpu (if available) in that core. If a cpu comes back online, then this cpu will be added to the core imc cpumask only if there was no other cpu from that core in the previous cpumask.

To register the hotplug functions for core_imc, a new state CPUHP_AP_PERF_POWERPC_COREIMC_ONLINE is added to the list of existing states.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h             |   1 +
 arch/powerpc/include/asm/opal-api.h            |  10 +-
 arch/powerpc/include/asm/opal.h                |   2 +
 arch/powerpc/perf/imc-pmu.c                    | 248 ++++++++++++++++++++++++-
 arch/powerpc/platforms/powernv/opal-imc.c      |   4 +-
 arch/powerpc/platforms/powernv/opal-wrappers.S |   1 +
 include/linux/cpuhotplug.h                     |   1 +
 7 files changed, 257 insertions(+), 10 deletions(-)
arch/powerpc/include/asm/imc-pmu.h +1 -0
index 59de083..5e76cd0 100644
@@ -21,6 +21,7 @@
 #define IMC_MAX_CHIPS			32
 #define IMC_MAX_PMUS			32
 #define IMC_MAX_PMU_NAME_LEN		256
+#define IMC_MAX_CORES			256
 
 #define NEST_IMC_ENGINE_START		1
 #define NEST_IMC_ENGINE_STOP		0
arch/powerpc/include/asm/opal-api.h +9 -1
index e15fb20..4ee52e8 100644
@@ -169,7 +169,8 @@
 #define OPAL_PCI_TCE_KILL			126
 #define OPAL_NMMU_SET_PTCR			127
 #define OPAL_NEST_IMC_COUNTERS_CONTROL		128
-#define OPAL_LAST				128
+#define OPAL_CORE_IMC_COUNTERS_CONTROL		129
+#define OPAL_LAST				129
 
 /* Device tree flags */
 
@@ -929,6 +930,13 @@ enum {
 	OPAL_PCI_TCE_KILL_ALL,
 };
 
+/* Operation argument to Core IMC */
+enum {
+	OPAL_CORE_IMC_DISABLE,
+	OPAL_CORE_IMC_ENABLE,
+	OPAL_CORE_IMC_INIT,
+};
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* __OPAL_API_H */
arch/powerpc/include/asm/opal.h +2 -0
index d93d082..c4baa6d 100644
@@ -229,6 +229,8 @@ int64_t opal_nmmu_set_ptcr(uint64_t chip_id, uint64_t ptcr);
 
 int64_t opal_nest_imc_counters_control(uint64_t mode, uint64_t value1,
 				uint64_t value2, uint64_t value3);
+int64_t opal_core_imc_counters_control(uint64_t operation, uint64_t addr,
+				uint64_t value2, uint64_t value3);
 
 /* Internal functions */
 extern int early_init_dt_scan_opal(unsigned long node, const char *uname,
arch/powerpc/perf/imc-pmu.c +241 -7

303 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal-imc.c +2 -2
index 67ce873..6db3c5f 100644
@@ -191,12 +191,12 @@ static int imc_events_node_parser(struct device_node *dev,
 				kfree(events[idx].ev_value);
 				continue;
 			}
+			idx++;
 			/*
 			 * If the common scale and unit properties available,
 			 * then, assign them to this event
 			 */
 			if (event_scale) {
-				idx++;
 				ret = set_event_property(event_scale, "scale",
 							 &events[idx],
 							 ev_name);
@@ -210,8 +210,8 @@ static int imc_events_node_parser(struct device_node *dev,
 							 ev_name);
 				if (ret)
 					continue;
+				idx++;
 			}
-			idx++;
 		} else if (strncmp(pp->name, "unit", 4) == 0) {
 			ret = set_event_property(pp, "unit", &events[idx],
 						 ev_name);
arch/powerpc/platforms/powernv/opal-wrappers.S +1 -0
index 9be9a9c..f8565d1 100644
@@ -302,3 +302,4 @@ OPAL_CALL(opal_int_set_mfrr,			OPAL_INT_SET_MFRR);
 OPAL_CALL(opal_pci_tce_kill,			OPAL_PCI_TCE_KILL);
 OPAL_CALL(opal_nmmu_set_ptcr,			OPAL_NMMU_SET_PTCR);
 OPAL_CALL(opal_nest_imc_counters_control,	OPAL_NEST_IMC_COUNTERS_CONTROL);
+OPAL_CALL(opal_core_imc_counters_control,	OPAL_CORE_IMC_COUNTERS_CONTROL);
include/linux/cpuhotplug.h +1 -0
index 8fac5ad..de4b90b 100644
@@ -139,6 +139,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_ARM_L2X0_ONLINE,
 	CPUHP_AP_PERF_ARM_QCOM_L2_ONLINE,
 	CPUHP_AP_PERF_POWERPC_NEST_ONLINE,
+	CPUHP_AP_PERF_POWERPC_COREIMC_ONLINE,
 	CPUHP_AP_WORKQUEUE_ONLINE,
 	CPUHP_AP_RCUTREE_ONLINE,
 	CPUHP_AP_ONLINE_DYN,
Signature · 1 line
2.7.4
#9 lore raw

PATCH v4 9/10 powerpc/powernv: Thread IMC events detection

In reply to Hemant Kumar

Patch adds support for detection of thread IMC events. It adds a new domain IMC_DOMAIN_THREAD and it is determined with the help of the compatibility string "ibm,imc-counters-thread" based on the IMC device tree.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h        |  2 ++
 arch/powerpc/perf/imc-pmu.c               |  1 +
 arch/powerpc/platforms/powernv/opal-imc.c | 11 +++++++++--
 3 files changed, 12 insertions(+), 2 deletions(-)
arch/powerpc/include/asm/imc-pmu.h +2 -0
index 5e76cd0..f2b4f12 100644
@@ -32,6 +32,7 @@
 #define IMC_DTB_COMPAT		"ibm,opal-in-memory-counters"
 #define IMC_DTB_NEST_COMPAT	"ibm,imc-counters-nest"
 #define IMC_DTB_CORE_COMPAT	"ibm,imc-counters-core"
+#define IMC_DTB_THREAD_COMPAT   "ibm,imc-counters-thread"
 
 /*
  * Structure to hold per chip specific memory address
@@ -70,6 +71,7 @@ struct imc_pmu {
  */
 #define IMC_DOMAIN_NEST		1
 #define IMC_DOMAIN_CORE		2
+#define IMC_DOMAIN_THREAD       3
 
 #define UNKNOWN_DOMAIN		-1
 
arch/powerpc/perf/imc-pmu.c +1 -0
index 61d99c7..a48c5be 100644
@@ -26,6 +26,7 @@ struct imc_pmu *core_imc_pmu;
 /* Needed for sanity check */
 extern u64 nest_max_offset;
 extern u64 core_max_offset;
+extern u64 thread_max_offset;
 
 PMU_FORMAT_ATTR(event, "config:0-20");
 static struct attribute *imc_format_attrs[] = {
arch/powerpc/platforms/powernv/opal-imc.c +9 -2
index 6db3c5f..a5565e7 100644
@@ -39,6 +39,7 @@ extern int init_imc_pmu(struct imc_events *events,
 			int idx, struct imc_pmu *pmu_ptr);
 u64 nest_max_offset;
 u64 core_max_offset;
+u64 thread_max_offset;
 
 static int imc_event_info(char *name, struct imc_events *events)
 {
@@ -86,6 +87,10 @@ static void update_max_value(u32 value, int pmu_domain)
 		if (core_max_offset < value)
 			core_max_offset = value;
 		break;
+	case IMC_DOMAIN_THREAD:
+		if (thread_max_offset < value)
+			thread_max_offset = value;
+		break;
 	default:
 		/* Unknown domain, return */
 		return;
@@ -239,6 +244,8 @@ int imc_get_domain(struct device_node *pmu_dev)
 		return IMC_DOMAIN_NEST;
 	if (of_device_is_compatible(pmu_dev, IMC_DTB_CORE_COMPAT))
 		return IMC_DOMAIN_CORE;
+	if (of_device_is_compatible(pmu_dev, IMC_DTB_THREAD_COMPAT))
+		return IMC_DOMAIN_THREAD;
 	else
 		return UNKNOWN_DOMAIN;
 }
@@ -277,7 +284,7 @@ static void imc_free_events(struct imc_events *events, int nr_entries)
 /*
  * imc_pmu_create : Takes the parent device which is the pmu unit and a
  *                  pmu_index as the inputs.
- * Allocates memory for the pmu, sets up its domain (NEST or CORE), and
+ * Allocates memory for the pmu, sets up its domain (NEST/CORE/THREAD), and
  * allocates memory for the events supported by this pmu. Assigns a name for
  * the pmu. Calls imc_events_node_parser() to setup the individual events.
  * If everything goes fine, it calls, init_imc_pmu() to setup the pmu device
@@ -305,7 +312,7 @@ static int imc_pmu_create(struct device_node *parent, int pmu_index)
 	if (pmu_ptr->domain == UNKNOWN_DOMAIN)
 		goto free_pmu;
 
-	/* Needed for hotplug/migration */
+	/* Needed for hotplug/migration for nest and core IMC PMUs */
 	if (pmu_ptr->domain == IMC_DOMAIN_CORE)
 		core_imc_pmu = pmu_ptr;
 	else if (pmu_ptr->domain == IMC_DOMAIN_NEST)
Signature · 1 line
2.7.4
#10 lore raw

PATCH v4 10/10 powerpc/perf: Thread IMC PMU functions

In reply to Hemant Kumar

This patch adds the PMU functions required for event initialization, read, update, add, del etc. for thread IMC PMU. Thread IMC PMUs are used for per-task monitoring. These PMUs don't need any hotplugging support.

For each CPU, a page of memory is allocated and is kept static i.e., these pages will exist till the machine shuts down. The base address of this page is assigned to the ldbar of that cpu. As soon as we do that, the thread IMC counters start running for that cpu and the data of these counters are assigned to the page allocated. But we use this for per-task monitoring. Whenever we start monitoring a task, the event is added is onto the task. At that point, we read the initial value of the event. Whenever, we stop monitoring the task, the final value is taken and the difference is the event data.

Now, a task can move to a different cpu. Suppose a task X is moving from cpu A to cpu B. When the task is scheduled out of A, we get an event_del for A, and hence, the event data is updated. And, we stop updating the X's event data. As soon as X moves on to B, event_add is called for B, and we again update the event_data. And this is how it keeps on updating the event data even when the task is scheduled on to different cpus.

Cc Madhavan Srinivasan <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Anton Blanchard <[email protected]>
Cc Sukadev Bhattiprolu <[email protected]>
Cc Michael Neuling <[email protected]>
Cc Stewart Smith <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Balbir Singh <[email protected]>
Cc Anju T Sudhakar <[email protected]>
---
 arch/powerpc/include/asm/imc-pmu.h |   4 +
 arch/powerpc/perf/imc-pmu.c        | 161 ++++++++++++++++++++++++++++++++++++-
 2 files changed, 164 insertions(+), 1 deletion(-)
arch/powerpc/include/asm/imc-pmu.h +4 -0
index f2b4f12..8b7141b 100644
@@ -22,6 +22,7 @@
 #define IMC_MAX_PMUS			32
 #define IMC_MAX_PMU_NAME_LEN		256
 #define IMC_MAX_CORES			256
+#define IMC_MAX_CPUS                    2048
 
 #define NEST_IMC_ENGINE_START		1
 #define NEST_IMC_ENGINE_STOP		0
@@ -34,6 +35,9 @@
 #define IMC_DTB_CORE_COMPAT	"ibm,imc-counters-core"
 #define IMC_DTB_THREAD_COMPAT   "ibm,imc-counters-thread"
 
+#define THREAD_IMC_LDBAR_MASK           0x0003ffffffffe000
+#define THREAD_IMC_ENABLE               0x8000000000000000
+
 /*
  * Structure to hold per chip specific memory address
  * information for nest pmus. Nest Counter data are exported
arch/powerpc/perf/imc-pmu.c +160 -1

212 lines not on this page · raw message

Signature · 1 line
2.7.4