* [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool
@ 2014-03-24 20:05 Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 1/2] power: Add idlestat tool for benchmarking energy-aware scheduler Zoran Markovic
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Zoran Markovic @ 2014-03-24 20:05 UTC (permalink / raw)
To: linux-kernel; +Cc: rob, mingo, peterz, rostedt, daniel.lezcano, zoran.markovic
Conclusions from Energy Aware Scheduling sessions at the latest Kernel Summit
identified a need for tools that would assess power consumption of the system
These tools would be used to prove efficiency of scheduler patches by
comparing power consumption before and after they were applied.
Attached is the proposal for the idlestat tool. The purpose of this patch
is to solicit feedback on tool's features, possible enhancements, etc.
Source code and sample idlestat report are provided for reference.
Please review and provide comments in anticipation of further development.
Regards, Zoran
Zoran Markovic (2):
power: Add idlestat tool for benchmarking energy-aware scheduler
sched: Add documentation for idlestat scheduler benchmarking tool
Documentation/scheduler/idlestat.txt | 79 +++
tools/power/idlestat/.gitignore | 50 ++
tools/power/idlestat/Makefile | 34 +
tools/power/idlestat/idlestat.c | 1229 ++++++++++++++++++++++++++++++++++
tools/power/idlestat/idlestat.h | 106 +++
tools/power/idlestat/list.h | 588 ++++++++++++++++
tools/power/idlestat/topology.c | 503 ++++++++++++++
tools/power/idlestat/topology.h | 77 +++
tools/power/idlestat/trace.c | 87 +++
tools/power/idlestat/trace.h | 43 ++
tools/power/idlestat/utils.c | 115 ++++
tools/power/idlestat/utils.h | 35 +
12 files changed, 2946 insertions(+)
create mode 100644 Documentation/scheduler/idlestat.txt
create mode 100644 tools/power/idlestat/.gitignore
create mode 100644 tools/power/idlestat/Makefile
create mode 100644 tools/power/idlestat/idlestat.c
create mode 100644 tools/power/idlestat/idlestat.h
create mode 100644 tools/power/idlestat/list.h
create mode 100644 tools/power/idlestat/topology.c
create mode 100644 tools/power/idlestat/topology.h
create mode 100644 tools/power/idlestat/trace.c
create mode 100644 tools/power/idlestat/trace.h
create mode 100644 tools/power/idlestat/utils.c
create mode 100644 tools/power/idlestat/utils.h
--
1.7.9.5
^ permalink raw reply [flat|nested] 4+ messages in thread
* [RFC PATCH 1/2] power: Add idlestat tool for benchmarking energy-aware scheduler
2014-03-24 20:05 [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool Zoran Markovic
@ 2014-03-24 20:05 ` Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 2/2] sched: Add documentation for idlestat scheduler benchmarking tool Zoran Markovic
2014-03-25 12:08 ` [RFC PATCH 0/2] sched: proposal " Preeti Murthy
2 siblings, 0 replies; 4+ messages in thread
From: Zoran Markovic @ 2014-03-24 20:05 UTC (permalink / raw)
To: linux-kernel; +Cc: rob, mingo, peterz, rostedt, daniel.lezcano, zoran.markovic
This is the initial snapshot of idlestat tool used to benchmark energy
efficiency of the scheduler. Functionality of this tool is described in
Documentation/power/idlestat.txt.
The code is still undergoing developement. This snapshot is provided for
reveiwers' reference with the intention of soliciting feedback on tool's
features.
Cc: Rob Landley <rob@landley.net>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Zoran Markovic <zoran.markovic@linaro.org>
---
tools/power/idlestat/.gitignore | 50 ++
tools/power/idlestat/Makefile | 34 ++
tools/power/idlestat/idlestat.c | 1229 +++++++++++++++++++++++++++++++++++++++
tools/power/idlestat/idlestat.h | 106 ++++
tools/power/idlestat/list.h | 588 +++++++++++++++++++
tools/power/idlestat/topology.c | 503 ++++++++++++++++
tools/power/idlestat/topology.h | 77 +++
tools/power/idlestat/trace.c | 87 +++
tools/power/idlestat/trace.h | 43 ++
tools/power/idlestat/utils.c | 115 ++++
tools/power/idlestat/utils.h | 35 ++
11 files changed, 2867 insertions(+)
create mode 100644 tools/power/idlestat/.gitignore
create mode 100644 tools/power/idlestat/Makefile
create mode 100644 tools/power/idlestat/idlestat.c
create mode 100644 tools/power/idlestat/idlestat.h
create mode 100644 tools/power/idlestat/list.h
create mode 100644 tools/power/idlestat/topology.c
create mode 100644 tools/power/idlestat/topology.h
create mode 100644 tools/power/idlestat/trace.c
create mode 100644 tools/power/idlestat/trace.h
create mode 100644 tools/power/idlestat/utils.c
create mode 100644 tools/power/idlestat/utils.h
diff --git a/tools/power/idlestat/.gitignore b/tools/power/idlestat/.gitignore
new file mode 100644
index 0000000..968abde
--- /dev/null
+++ b/tools/power/idlestat/.gitignore
@@ -0,0 +1,50 @@
+#
+# NOTE! Don't add files that are generated in specific
+# subdirectories here. Add them in the ".gitignore" file
+# in that subdirectory instead.
+#
+# NOTE! Please use 'git ls-files -i --exclude-standard'
+# command after changing this file, to see if there are
+# any tracked files which get ignored after the change.
+#
+# Normal rules
+#
+.*
+*.o
+*.o.*
+*.a
+*.s
+*.so
+*.so.dbg
+*.i
+*.elf
+*.bin
+*.gz
+*.bz2
+*.lzma
+*.xz
+*.lz4
+*.lzo
+*.patch
+*.gcno
+*.orig
+*~
+\#*#
+
+#
+# Top-level files
+#
+/idlestat
+
+#
+# git files that we don't want to ignore even it they are dot-files
+#
+!.gitignore
+!.mailmap
+
+# gnu global files
+GPATH
+GRTAGS
+GSYMS
+GTAGS
+
diff --git a/tools/power/idlestat/Makefile b/tools/power/idlestat/Makefile
new file mode 100644
index 0000000..e351b4d
--- /dev/null
+++ b/tools/power/idlestat/Makefile
@@ -0,0 +1,34 @@
+#
+# Makefile
+#
+# Copyright (C) 2014, Linaro Limited
+#
+# 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.
+#
+# Contributors:
+# Daniel Lezcano <daniel.lezcano@linaro.org>
+# Zoran Markovic <zoran.markovic@linaro.org>
+#
+CFLAGS?=-g -Wall
+CC?=gcc
+
+OBJS = idlestat.o topology.o trace.o utils.o
+
+default: idlestat
+
+idlestat: $(OBJS)
+ $(CC) ${CFLAGS} $(OBJS) -o $@
+
+clean:
+ rm -f $(OBJS) idlestat
diff --git a/tools/power/idlestat/idlestat.c b/tools/power/idlestat/idlestat.c
new file mode 100644
index 0000000..074926e
--- /dev/null
+++ b/tools/power/idlestat/idlestat.c
@@ -0,0 +1,1229 @@
+/*
+ * idlestat.c
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#define _GNU_SOURCE
+#include <getopt.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <unistd.h>
+#include <sched.h>
+#include <string.h>
+#include <float.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/resource.h>
+#include <assert.h>
+
+#include "idlestat.h"
+#include "utils.h"
+#include "trace.h"
+#include "list.h"
+#include "topology.h"
+
+#define IDLESTAT_VERSION "0.3-rc1"
+
+static char irq_type_name[][8] = {
+ "irq",
+ "ipi",
+ };
+
+static char buffer[BUFSIZE];
+
+static inline int error(const char *str)
+{
+ perror(str);
+ return -1;
+}
+
+static inline void *ptrerror(const char *str)
+{
+ perror(str);
+ return NULL;
+}
+
+static int dump_states(struct cpuidle_cstates *cstates,
+ struct cpufreq_pstates *pstates,
+ int state, int count, char *str)
+{
+ int j, k;
+ struct cpuidle_cstate *cstate;
+
+ for (j = 0; j < cstates->cstate_max + 1; j++) {
+
+ if (state != -1 && state != j)
+ continue;
+
+ cstate = &cstates->cstate[j];
+
+ for (k = 0; k < MIN(count, cstate->nrdata); k++) {
+ printf("%lf %d\n", cstate->data[k].begin, j);
+ printf("%lf 0\n", cstate->data[k].end);
+ }
+
+ /* add a break */
+ printf("\n");
+ }
+
+ return 0;
+}
+
+static int display_states(struct cpuidle_cstates *cstates,
+ struct cpufreq_pstates *pstates,
+ int state, int count, char *str)
+{
+ int j;
+
+ printf("%s@state\thits\t total(us)\t\tavg(us)\tmin(us)\t"
+ "max(us)\n", str);
+ for (j = 0; j < cstates->cstate_max + 1; j++) {
+ struct cpuidle_cstate *c = &cstates->cstate[j];
+
+ if (state != -1 && state != j)
+ continue;
+
+ printf("%*c %s\t%d\t%15.2lf\t%15.2lf\t%.2lf\t%.2lf\n",
+ (int)strlen(str), ' ',
+ c->name, c->nrdata, c->duration,
+ c->avg_time,
+ (c->min_time == DBL_MAX ? 0. : c->min_time),
+ c->max_time);
+ }
+ if (pstates) {
+ for (j = 0; j < pstates->max; j++) {
+ struct cpufreq_pstate *p = &(pstates->pstate[j]);
+ printf("%*c %d\t%d\t%15.2lf\t%15.2lf\t%.2lf\t%.2lf\n",
+ (int)strlen(str), ' ',
+ p->freq/1000, p->count, p->duration,
+ p->avg_time,
+ (p->min_time == DBL_MAX ? 0. : p->min_time),
+ p->max_time);
+ }
+ }
+
+ if (strstr(str, IRQ_WAKEUP_UNIT_NAME)) {
+ struct wakeup_info *wakeinfo = &cstates->wakeinfo;
+ struct wakeup_irq *irqinfo = wakeinfo->irqinfo;
+ printf("%s wakeups \tname \t\tcount\n", str);
+ for (j = 0; j < wakeinfo->nrdata; j++, irqinfo++) {
+ printf("%*c %s%03d\t%-15.15s\t%d\n", (int)strlen(str),
+ ' ',
+ (irqinfo->irq_type < IRQ_TYPE_MAX) ?
+ irq_type_name[irqinfo->irq_type] : "NULL",
+ irqinfo->id, irqinfo->name, irqinfo->count);
+ }
+ }
+
+ return 0;
+}
+
+int dump_all_data(struct cpuidle_datas *datas, int state, int count,
+ int (*dump)(struct cpuidle_cstates *,
+ struct cpufreq_pstates *, int, int, char *))
+{
+ int i = 0, nrcpus = datas->nrcpus;
+ struct cpuidle_cstates *cstates;
+ struct cpufreq_pstates *pstates;
+
+ do {
+ cstates = &datas->cstates[i];
+ pstates = &datas->pstates[i];
+
+ if (nrcpus == -1)
+ sprintf(buffer, "cluster");
+ else
+ sprintf(buffer, "cpu%d", i);
+
+ dump(cstates, pstates, state, count, buffer);
+
+ i++;
+
+ } while (i < nrcpus && nrcpus != -1);
+
+ return 0;
+}
+
+static struct cpuidle_data *intersection(struct cpuidle_data *data1,
+ struct cpuidle_data *data2)
+{
+ double begin, end;
+ struct cpuidle_data *data;
+
+ begin = MAX(data1->begin, data2->begin);
+ end = MIN(data1->end, data2->end);
+
+ if (begin >= end)
+ return NULL;
+
+ data = malloc(sizeof(*data));
+ if (!data)
+ return NULL;
+
+ data->begin = begin;
+ data->end = end;
+ data->duration = end - begin;
+ data->duration *= 1000000;
+
+ return data;
+}
+
+static struct cpuidle_cstate *inter(struct cpuidle_cstate *c1,
+ struct cpuidle_cstate *c2)
+{
+ int i, j;
+ struct cpuidle_data *interval;
+ struct cpuidle_cstate *result;
+ struct cpuidle_data *data = NULL;
+ size_t index;
+
+ if (!c1)
+ return c2;
+ if (!c2)
+ return c1;
+
+ result = calloc(sizeof(*result), 1);
+ if (!result)
+ return NULL;
+
+ for (i = 0, index = 0; i < c1->nrdata; i++) {
+
+ for (j = index; j < c2->nrdata; j++) {
+ struct cpuidle_data *tmp;
+
+ /* intervals are ordered, no need to go further */
+ if (c1->data[i].end < c2->data[j].begin)
+ break;
+
+ /* primary loop begins where we ended */
+ if (c1->data[i].begin > c2->data[j].end)
+ index = j;
+
+ interval = intersection(&c1->data[i], &c2->data[j]);
+ if (!interval)
+ continue;
+
+ result->min_time = MIN(result->min_time,
+ interval->duration);
+
+ result->max_time = MAX(result->max_time,
+ interval->duration);
+
+ result->avg_time = AVG(result->avg_time,
+ interval->duration,
+ result->nrdata + 1);
+
+ result->duration += interval->duration;
+
+ result->nrdata++;
+
+ tmp = realloc(data, sizeof(*data) *
+ (result->nrdata + 1));
+ if (!tmp) {
+ free(data);
+ free(result);
+ return NULL;
+ }
+ data = tmp;
+
+ result->data = data;
+ result->data[result->nrdata - 1] = *interval;
+
+ free(interval);
+ }
+ }
+
+ return result;
+}
+
+#define CPUIDLE_STATENAME_PATH_FORMAT \
+ "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name"
+
+static char *cpuidle_cstate_name(int cpu, int state)
+{
+ char *fpath, *name;
+ FILE *snf;
+ char line[256];
+
+ if (asprintf(&fpath, CPUIDLE_STATENAME_PATH_FORMAT, cpu, state) < 0)
+ return NULL;
+
+ /* read cpuidle state name for the CPU */
+ snf = fopen(fpath, "r");
+ if (!snf) {
+ free(fpath);
+ return NULL;
+ }
+
+ name = fgets(line, sizeof(line)/sizeof(line[0]), snf);
+ if (!name)
+ goto free_exit;
+
+ /* get rid of trailing characters and duplicate string */
+ name = strtok(name, "\n ");
+ name = strdup(name);
+
+free_exit:
+ fclose(snf);
+ free(fpath);
+ return name;
+}
+
+
+/**
+ * release_cstate_info - free all C-state related structs
+ * @cstates: per-cpu array of C-state statistics structs
+ * @nrcpus: number of CPUs
+ */
+static void release_cstate_info(struct cpuidle_cstates *cstates, int nrcpus)
+{
+ int cpu, i;
+
+ if (!cstates)
+ /* already cleaned up */
+ return;
+
+ /* free C-state names */
+ for (cpu = 0; cpu < nrcpus; cpu++) {
+ for (i = 0; i < MAXCSTATE; i++) {
+ struct cpuidle_cstate *c = &(cstates[cpu].cstate[i]);
+ if (c->name)
+ free(c->name);
+ }
+ }
+
+ /* free the cstates array */
+ free(cstates);
+}
+
+/**
+ * build_cstate_info - parse cpuidle sysfs entries and build per-CPU
+ * structs to maintain statistics of C-state transitions
+ * @nrcpus: number of CPUs
+ *
+ * Return: per-CPU array of structs (success) or NULL (error)
+ */
+static struct cpuidle_cstates *build_cstate_info(int nrcpus)
+{
+ int cpu;
+ struct cpuidle_cstates *cstates;
+
+ cstates = calloc(nrcpus, sizeof(*cstates));
+ if (!cstates)
+ return NULL;
+ memset(cstates, 0, sizeof(*cstates) * nrcpus);
+
+ /* initialize cstate_max for each cpu */
+ for (cpu = 0; cpu < nrcpus; cpu++) {
+ int i;
+ struct cpuidle_cstate *c;
+ cstates[cpu].cstate_max = -1;
+ cstates[cpu].last_cstate = -1;
+ for (i = 0; i < MAXCSTATE; i++) {
+ c = &(cstates[cpu].cstate[i]);
+ c->name = cpuidle_cstate_name(cpu, i);
+ c->data = NULL;
+ c->nrdata = 0;
+ c->avg_time = 0.;
+ c->max_time = 0.;
+ c->min_time = DBL_MAX;
+ c->duration = 0.;
+ }
+ }
+ return cstates;
+}
+
+#define CPUFREQ_AVFREQ_PATH_FORMAT \
+ "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_available_frequencies"
+
+/**
+ * release_pstate_info - free all P-state related structs
+ * @pstates: per-cpu array of P-state statistics structs
+ * @nrcpus: number of CPUs
+ */
+static void release_pstate_info(struct cpufreq_pstates *pstates, int nrcpus)
+{
+ int cpu;
+
+ if (!pstates)
+ /* already cleaned up */
+ return;
+
+ /* first check and clean per-cpu structs */
+ for (cpu = 0; cpu < nrcpus; cpu++)
+ if (pstates[cpu].pstate)
+ free(pstates[cpu].pstate);
+
+ /* now free the master cpufreq structs */
+ free(pstates);
+
+ return;
+}
+
+/**
+ * build_pstate_info - parse cpufreq sysfs entries and build per-CPU
+ * structs to maintain statistics of P-state transitions
+ * @nrcpus: number of CPUs
+ *
+ * Return: per-CPU array of structs (success) or NULL (error)
+ */
+static struct cpufreq_pstates *build_pstate_info(int nrcpus)
+{
+ int cpu;
+ struct cpufreq_pstates *pstates;
+
+ pstates = calloc(nrcpus, sizeof(*pstates));
+ if (!pstates)
+ return NULL;
+ memset(pstates, 0, sizeof(*pstates) * nrcpus);
+
+ for (cpu = 0; cpu < nrcpus; cpu++) {
+ struct cpufreq_pstate *pstate;
+ int nrfreq;
+ char *fpath, *freq, line[256];
+ FILE *sc_av_freq;
+
+ if (asprintf(&fpath, CPUFREQ_AVFREQ_PATH_FORMAT, cpu) < 0)
+ goto clean_exit;
+
+ /* read scaling_available_frequencies for the CPU */
+ sc_av_freq = fopen(fpath, "r");
+ free(fpath);
+ if (!sc_av_freq)
+ goto clean_exit;
+ freq = fgets(line, sizeof(line)/sizeof(line[0]), sc_av_freq);
+ fclose(sc_av_freq);
+ if (!freq)
+ goto clean_exit;
+
+ /* tokenize line and populate each frequency */
+ nrfreq = 0;
+ pstate = NULL;
+ while ((freq = strtok(freq, "\n ")) != NULL) {
+ pstate = realloc(pstate, sizeof(*pstate) * (nrfreq+1));
+ if (!pstate)
+ goto clean_exit;
+
+ /* initialize pstate record */
+ pstate[nrfreq].id = nrfreq;
+ pstate[nrfreq].freq = atol(freq);
+ pstate[nrfreq].count = 0;
+ pstate[nrfreq].min_time = DBL_MAX;
+ pstate[nrfreq].max_time = 0.;
+ pstate[nrfreq].avg_time = 0.;
+ pstate[nrfreq].duration = 0.;
+ nrfreq++;
+ freq = NULL;
+ }
+
+ /* now populate cpufreq_pstates for this CPU */
+ pstates[cpu].pstate = pstate;
+ pstates[cpu].max = nrfreq;
+ pstates[cpu].current = -1; /* unknown */
+ pstates[cpu].idle = -1; /* unknown */
+ pstates[cpu].time_enter = 0.;
+ pstates[cpu].time_exit = 0.;
+ }
+
+ return pstates;
+
+clean_exit:
+ release_pstate_info(pstates, nrcpus);
+ return NULL;
+}
+
+static int get_current_pstate(struct cpuidle_datas *datas, int cpu,
+ struct cpufreq_pstates **pstates,
+ struct cpufreq_pstate **pstate)
+{
+ struct cpufreq_pstates *ps;
+
+ if (cpu < 0 || cpu > datas->nrcpus)
+ return -2;
+
+ ps = &(datas->pstates[cpu]);
+
+ *pstate = (ps->current == -1 ? NULL : &(ps->pstate[ps->current]));
+ *pstates = ps;
+
+ /* return 1 if CPU is idle, otherwise return 0 */
+ return ps->idle;
+}
+
+static int freq_to_pstate_index(struct cpufreq_pstates *ps, int freq)
+{
+ int i;
+
+ /* find frequency in table of P-states */
+ for (i = 0; i < ps->max && freq != ps->pstate[i].freq; i++)
+ /* just search */;
+
+ /* if not found, return -1 */
+ return i >= ps->max ? -1 : ps->pstate[i].id;
+}
+
+static void open_current_pstate(struct cpufreq_pstates *ps, double time)
+{
+ ps->time_enter = time;
+}
+
+static void open_next_pstate(struct cpufreq_pstates *ps, int s, double time)
+{
+ ps->current = s;
+ if (ps->idle) {
+ fprintf(stderr, "warning: opening P-state on idle CPU\n");
+ return;
+ }
+ open_current_pstate(ps, time);
+}
+
+#define USEC_PER_SEC 1000000
+static void close_current_pstate(struct cpufreq_pstates *ps, double time)
+{
+ int c = ps->current;
+ struct cpufreq_pstate *p = &(ps->pstate[c]);
+ double elapsed;
+
+ if (ps->idle) {
+ fprintf(stderr, "warning: closing P-state on idle CPU\n");
+ return;
+ }
+ elapsed = (time - ps->time_enter) * USEC_PER_SEC;
+ p->min_time = MIN(p->min_time, elapsed);
+ p->max_time = MAX(p->max_time, elapsed);
+ p->avg_time = AVG(p->avg_time, elapsed, p->count + 1);
+ p->duration += elapsed;
+ p->count++;
+}
+
+static void cpu_change_pstate(struct cpuidle_datas *datas, int cpu,
+ int freq, double time)
+{
+ struct cpufreq_pstates *ps;
+ struct cpufreq_pstate *p;
+ int cur, next;
+
+ cur = get_current_pstate(datas, cpu, &ps, &p);
+ next = freq_to_pstate_index(ps, freq);
+
+ switch (cur) {
+ case 1:
+ /* if CPU is idle, update current state and leave
+ * stats unchanged
+ */
+ ps->current = next;
+ return;
+
+ case -1:
+ /* current pstate is -1, i.e. this is the first update */
+ open_next_pstate(ps, next, time);
+ return;
+
+ case 0:
+ /* running CPU, update all stats, but skip closing current
+ * state if it's the initial update for CPU
+ */
+ if (p)
+ close_current_pstate(ps, time);
+ open_next_pstate(ps, next, time);
+ return;
+
+ default:
+ fprintf(stderr, "illegal pstate %d for cpu %d, exiting.\n",
+ cur, cpu);
+ exit(-1);
+ }
+}
+
+static void cpu_pstate_idle(struct cpuidle_datas *datas, int cpu, double time)
+{
+ struct cpufreq_pstates *ps = &(datas->pstates[cpu]);
+ if (ps->current != -1)
+ close_current_pstate(ps, time);
+ ps->idle = 1;
+}
+
+static void cpu_pstate_running(struct cpuidle_datas *datas, int cpu,
+ double time)
+{
+ struct cpufreq_pstates *ps = &(datas->pstates[cpu]);
+ ps->idle = 0;
+ if (ps->current != -1)
+ open_current_pstate(ps, time);
+}
+
+static int store_data(double time, int state, int cpu,
+ struct cpuidle_datas *datas, int count)
+{
+ struct cpuidle_cstates *cstates = &datas->cstates[cpu];
+ struct cpuidle_cstate *cstate;
+ struct cpuidle_data *data, *tmp;
+ int nrdata, last_cstate = cstates->last_cstate;
+
+ /* ignore when we got a "closing" state first */
+ if (state == -1 && cstates->cstate_max == -1)
+ return 0;
+
+ cstate = &cstates->cstate[state == -1 ? last_cstate : state];
+ data = cstate->data;
+ nrdata = cstate->nrdata;
+
+ if (state == -1) {
+
+ data = &data[nrdata];
+
+ data->end = time;
+ data->duration = data->end - data->begin;
+
+ /* That happens when precision digit in the file exceed
+ * 7 (eg. xxx.1000000). Ignoring the result because I don't
+ * find a way to fix with the sscanf used in the caller
+ */
+ if (data->duration < 0)
+ return 0;
+
+ /* convert to us */
+ data->duration *= 1000000;
+
+ cstate->min_time = MIN(cstate->min_time, data->duration);
+
+ cstate->max_time = MAX(cstate->max_time, data->duration);
+
+
+ cstate->avg_time = AVG(cstate->avg_time, data->duration,
+ cstate->nrdata + 1);
+
+ cstate->duration += data->duration;
+
+ cstate->nrdata++;
+
+ /* need indication if CPU is idle or not */
+ cstates->last_cstate = -1;
+ cpu_pstate_running(datas, cpu, time);
+
+ return 0;
+ }
+
+ tmp = realloc(data, sizeof(*data) * (nrdata + 1));
+ if (!tmp) {
+ free(data);
+ return error("realloc data");
+ }
+ data = tmp;
+
+ data[nrdata].begin = time;
+
+ cstates->cstate[state].data = data;
+ cstates->cstate_max = MAX(cstates->cstate_max, state);
+ cstates->last_cstate = state;
+ cstates->wakeirq = NULL;
+ cpu_pstate_idle(datas, cpu, time);
+
+ return 0;
+}
+
+static struct wakeup_irq *find_irqinfo(struct wakeup_info *wakeinfo, int irqid)
+{
+ struct wakeup_irq *irqinfo;
+ int i;
+
+ for (i = 0; i < wakeinfo->nrdata; i++) {
+ irqinfo = &wakeinfo->irqinfo[i];
+ if (irqinfo->id == irqid)
+ return irqinfo;
+ }
+
+ return NULL;
+}
+
+static int store_irq(int cpu, int irqid, char *irqname,
+ struct cpuidle_datas *datas, int count, int irq_type)
+{
+ struct cpuidle_cstates *cstates = &datas->cstates[cpu];
+ struct wakeup_irq *irqinfo;
+ struct wakeup_info *wakeinfo = &cstates->wakeinfo;
+
+ if (cstates->wakeirq != NULL)
+ return 0;
+
+ irqinfo = find_irqinfo(wakeinfo, irqid);
+ if (NULL == irqinfo) {
+ irqinfo = realloc(wakeinfo->irqinfo,
+ sizeof(*irqinfo) * (wakeinfo->nrdata + 1));
+ if (!irqinfo)
+ return error("realloc irqinfo");
+
+ wakeinfo->irqinfo = irqinfo;
+
+ irqinfo = &wakeinfo->irqinfo[wakeinfo->nrdata++];
+ irqinfo->id = irqid;
+ strcpy(irqinfo->name, irqname);
+ irqinfo->irq_type = irq_type;
+ irqinfo->count = 0;
+ }
+
+ irqinfo->count++;
+
+ cstates->wakeirq = irqinfo;
+
+ return 0;
+}
+
+#define TRACE_IRQ_FORMAT "%*[^[][%d] %*[^=]=%d%*[^=]=%16s"
+#define TRACE_IPIIRQ_FORMAT "%*[^[][%d] %*[^=]=%d%*[^=]=%16s"
+
+#define TRACE_CMD_FORMAT "%*[^]]] %lf:%*[^=]=%u%*[^=]=%d"
+#define TRACE_FORMAT "%*[^]]] %*s %lf:%*[^=]=%u%*[^=]=%d"
+
+static int get_wakeup_irq(struct cpuidle_datas *datas, char *buffer, int count)
+{
+ int cpu, irqid;
+ char irqname[NAMELEN+1];
+
+ if (strstr(buffer, "irq_handler_entry")) {
+ assert(sscanf(buffer, TRACE_IRQ_FORMAT, &cpu, &irqid,
+ irqname) == 3);
+
+ store_irq(cpu, irqid, irqname, datas, count, HARD_IRQ);
+ return 0;
+ }
+
+ if (strstr(buffer, "ipi_handler_entry")) {
+ assert(sscanf(buffer, TRACE_IPIIRQ_FORMAT, &cpu, &irqid,
+ irqname) == 3);
+
+ store_irq(cpu, irqid, irqname, datas, count, IPI_IRQ);
+ return 0;
+ }
+
+ return -1;
+}
+
+static struct cpuidle_datas *idlestat_load(const char *path)
+{
+ FILE *f;
+ unsigned int state = 0, freq = 0, cpu = 0, nrcpus = 0;
+ double time, begin = 0, end = 0;
+ size_t count, start = 1;
+ struct cpuidle_datas *datas;
+ int ret;
+
+ f = fopen(path, "r");
+ if (!f)
+ return ptrerror("fopen");
+
+ /* version line */
+ fgets(buffer, BUFSIZE, f);
+
+ fgets(buffer, BUFSIZE, f);
+ assert(sscanf(buffer, "cpus=%u", &nrcpus) == 1);
+
+ if (!nrcpus) {
+ fclose(f);
+ return ptrerror("read error for 'cpus=' in trace file");
+ }
+
+ datas = malloc(sizeof(*datas));
+ if (!datas) {
+ fclose(f);
+ return ptrerror("malloc datas");
+ }
+
+ datas->cstates = build_cstate_info(nrcpus);
+ if (!datas->cstates) {
+ free(datas);
+ fclose(f);
+ return ptrerror("calloc cstate");
+ }
+
+ datas->pstates = build_pstate_info(nrcpus);
+ if (!datas->pstates)
+ return ptrerror("calloc pstate");
+
+ datas->nrcpus = nrcpus;
+
+ fgets(buffer, BUFSIZE, f);
+
+ /* read topology information */
+ read_cpu_topo_info(f, buffer);
+
+ do {
+ if (strstr(buffer, "cpu_idle")) {
+ assert(sscanf(buffer, TRACE_FORMAT, &time, &state,
+ &cpu) == 3);
+
+ if (start) {
+ begin = time;
+ start = 0;
+ }
+ end = time;
+
+ store_data(time, state, cpu, datas, count);
+ count++;
+ continue;
+ } else if (strstr(buffer, "cpu_frequency")) {
+ assert(sscanf(buffer, TRACE_FORMAT, &time, &freq,
+ &cpu) == 3);
+ cpu_change_pstate(datas, cpu, freq, time);
+ continue;
+ }
+
+ ret = get_wakeup_irq(datas, buffer, count);
+ count += (0 == ret) ? 1 : 0;
+
+ } while (fgets(buffer, BUFSIZE, f));
+
+ fclose(f);
+
+ fprintf(stderr, "Log is %lf secs long with %zd events\n",
+ end - begin, count);
+
+ return datas;
+}
+
+struct cpuidle_datas *cluster_data(struct cpuidle_datas *datas)
+{
+ struct cpuidle_cstate *c1, *cstates;
+ struct cpuidle_datas *result;
+ int i, j;
+ int cstate_max = -1;
+
+ result = malloc(sizeof(*result));
+ if (!result)
+ return NULL;
+
+ result->nrcpus = -1; /* the cluster */
+
+ result->cstates = calloc(sizeof(*result->cstates), 1);
+ if (!result->cstates)
+ return NULL;
+
+ /* hack but negligeable overhead */
+ for (i = 0; i < datas->nrcpus; i++)
+ cstate_max = MAX(cstate_max, datas->cstates[i].cstate_max);
+ result->cstates[0].cstate_max = cstate_max;
+
+ for (i = 0; i < cstate_max + 1; i++) {
+
+ for (j = 0, cstates = NULL; j < datas->nrcpus; j++) {
+
+ c1 = &datas->cstates[j].cstate[i];
+
+ cstates = inter(cstates, c1);
+ if (!cstates)
+ continue;
+ }
+
+ /* copy state names from the first cpu */
+ cstates->name = strdup(datas->cstates[0].cstate[i].name);
+
+ result->cstates[0].cstate[i] = *cstates;
+ }
+
+ return result;
+}
+
+struct cpuidle_cstates *core_cluster_data(struct cpu_core *s_core)
+{
+ struct cpuidle_cstate *c1, *cstates;
+ struct cpuidle_cstates *result;
+ struct cpu_cpu *s_cpu;
+ int i;
+ int cstate_max = -1;
+
+ if (!s_core->is_ht)
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu)
+ return s_cpu->cstates;
+
+ result = calloc(sizeof(*result), 1);
+ if (!result)
+ return NULL;
+
+ /* hack but negligeable overhead */
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu)
+ cstate_max = MAX(cstate_max, s_cpu->cstates->cstate_max);
+ result->cstate_max = cstate_max;
+
+ for (i = 0; i < cstate_max + 1; i++) {
+ cstates = NULL;
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu) {
+ c1 = &s_cpu->cstates->cstate[i];
+
+ cstates = inter(cstates, c1);
+ if (!cstates)
+ continue;
+ }
+ /* copy state name from first cpu */
+ s_cpu = list_first_entry(&s_core->cpu_head, struct cpu_cpu,
+ list_cpu);
+ cstates->name = strdup(s_cpu->cstates->cstate[i].name);
+
+ result->cstate[i] = *cstates;
+ }
+
+ return result;
+}
+
+struct cpuidle_cstates *physical_cluster_data(struct cpu_physical *s_phy)
+{
+ struct cpuidle_cstate *c1, *cstates;
+ struct cpuidle_cstates *result;
+ struct cpu_core *s_core;
+ int i;
+ int cstate_max = -1;
+
+ result = calloc(sizeof(*result), 1);
+ if (!result)
+ return NULL;
+
+ /* hack but negligeable overhead */
+ list_for_each_entry(s_core, &s_phy->core_head, list_core)
+ cstate_max = MAX(cstate_max, s_core->cstates->cstate_max);
+ result->cstate_max = cstate_max;
+
+ for (i = 0; i < cstate_max + 1; i++) {
+ cstates = NULL;
+ list_for_each_entry(s_core, &s_phy->core_head, list_core) {
+ c1 = &s_core->cstates->cstate[i];
+
+ cstates = inter(cstates, c1);
+ if (!cstates)
+ continue;
+ }
+ /* copy state name from first core */
+ s_core = list_first_entry(&s_phy->core_head, struct cpu_core,
+ list_core);
+ cstates->name = strdup(s_core->cstates->cstate[i].name);
+
+ result->cstate[i] = *cstates;
+ }
+
+ return result;
+}
+
+static void help(const char *cmd)
+{
+ fprintf(stderr,
+ "%s [-d|--dump] [-c|--cstate=x] [-o|--output-file] <file>\n",
+ basename(cmd));
+}
+
+static void version(const char *cmd)
+{
+ printf("%s version %s\n", basename(cmd), IDLESTAT_VERSION);
+}
+
+static struct option long_options[] = {
+ { "dump", 0, 0, 'd' },
+ { "iterations", 0, 0, 'i' },
+ { "cstate", 0, 0, 'c' },
+ { "debug", 0, 0, 'g' },
+ { "output-file", 0, 0, 'o' },
+ { "verbose", 0, 0, 'v' },
+ { "version", 0, 0, 'V' },
+ { "help", 0, 0, 'h' },
+ { 0, 0, 0, 0 }
+};
+
+struct idledebug_options {
+ bool debug;
+ bool dump;
+ int cstate;
+ int iterations;
+ char *filename;
+ unsigned int duration;
+};
+
+int getoptions(int argc, char *argv[], struct idledebug_options *options)
+{
+ int c;
+
+ memset(options, 0, sizeof(*options));
+ options->cstate = -1;
+ options->filename = NULL;
+
+ while (1) {
+
+ int optindex = 0;
+
+ c = getopt_long(argc, argv, "gdvVho:i:c:t:",
+ long_options, &optindex);
+ if (c == -1)
+ break;
+
+ switch (c) {
+ case 'g':
+ options->debug = true;
+ break;
+ case 'd':
+ options->dump = true;
+ break;
+ case 'i':
+ options->iterations = atoi(optarg);
+ break;
+ case 'c':
+ options->cstate = atoi(optarg);
+ break;
+ case 't':
+ options->duration = atoi(optarg);
+ break;
+ case 'o':
+ options->filename = optarg;
+ break;
+ case 'h':
+ help(argv[0]);
+ exit(0);
+ break;
+ case 'V':
+ version(argv[0]);
+ exit(0);
+ break;
+ case '?':
+ fprintf(stderr, "%s: Unknown option %c'.\n",
+ argv[0], optopt);
+ /* fall through */
+ default:
+ return -1;
+ }
+ }
+
+ if (options->cstate >= MAXCSTATE) {
+ fprintf(stderr, "C-state must be less than %d\n", MAXCSTATE);
+ return -1;
+ }
+
+ if (options->iterations < 0)
+ fprintf(stderr, "dump values must be a positive value\n");
+
+ if (NULL == options->filename) {
+ fprintf(stderr, "expected filename\n");
+ return -1;
+ }
+
+ return 0;
+}
+
+static int idlestat_file_for_each_line(const char *path, void *data,
+ int (*handler)(const char *, void *))
+{
+ FILE *f;
+ int ret;
+
+ if (!handler)
+ return -1;
+
+ f = fopen(path, "r");
+
+ if (!f) {
+ fprintf(stderr, "failed to open '%s': %m\n", path);
+ return -1;
+ }
+
+ while (fgets(buffer, BUFSIZE, f)) {
+ ret = handler(buffer, data);
+ if (ret)
+ break;
+ }
+
+ fclose(f);
+
+ return ret;
+}
+
+static int idlestat_store(const char *path)
+{
+ FILE *f;
+ int ret;
+
+ ret = sysconf(_SC_NPROCESSORS_CONF);
+ if (ret < 0)
+ return -1;
+
+ f = fopen(path, "w+");
+ if (!f) {
+ fprintf(f, "failed to open '%s': %m\n", path);
+ return -1;
+ }
+
+ fprintf(f, "version = 1\n");
+ fprintf(f, "cpus=%d\n", ret);
+
+ /* output topology information */
+ output_cpu_topo_info(f);
+
+ ret = idlestat_file_for_each_line(TRACE_FILE, f, store_line);
+
+ fclose(f);
+
+ return ret;
+}
+
+static int idlestat_wake_all(void)
+{
+ int rcpu, i, ret;
+ cpu_set_t cpumask;
+
+ ret = sysconf(_SC_NPROCESSORS_CONF);
+ if (ret < 0)
+ return -1;
+
+ rcpu = sched_getcpu();
+ if (rcpu < 0)
+ return -1;
+
+ for (i = 0; i < ret; i++) {
+
+ /* Pointless to wake up ourself */
+ if (i == rcpu)
+ continue;
+
+ CPU_ZERO(&cpumask);
+ CPU_SET(i, &cpumask);
+
+ sched_setaffinity(0, sizeof(cpumask), &cpumask);
+ }
+
+ return 0;
+}
+
+int main(int argc, char *argv[])
+{
+ struct cpuidle_datas *datas;
+ struct cpuidle_datas *cluster;
+ struct idledebug_options options;
+ struct rusage rusage;
+
+ if (getoptions(argc, argv, &options))
+ return 1;
+
+ /* We have to manipulate some files only accessible to root */
+ if (getuid()) {
+ fprintf(stderr, "must be root to run the tool\n");
+ return -1;
+ }
+
+ /* init cpu topoinfo */
+ init_cpu_topo_info();
+
+ /* Acquisition time specified means we will get the traces */
+ if (options.duration) {
+
+ /* Read cpu topology info from sysfs */
+ read_sysfs_cpu_topo();
+
+ /* Stop tracing (just in case) */
+ if (idlestat_trace_enable(false))
+ return -1;
+
+ /* Initialize the traces for cpu_idle and increase the
+ * buffer size to let 'idlestat' to sleep instead of
+ * acquiring data, hence preventing it to pertubate the
+ * measurements. */
+ if (idlestat_init_trace(options.duration))
+ return 1;
+
+ /* Remove all the previous traces */
+ if (idlestat_flush_trace())
+ return -1;
+
+ /* Start the recording */
+ if (idlestat_trace_enable(true))
+ return -1;
+ /* We want to prevent to begin the acquisition with a cpu in
+ * idle state because we won't be able later to close the
+ * state and to determine which state it was. */
+ if (idlestat_wake_all())
+ return -1;
+
+ /* Do nothing */
+ sleep(options.duration);
+
+ /* Wake up all cpus again to account for last idle state */
+ if (idlestat_wake_all())
+ return -1;
+
+ /* Stop tracing */
+ if (idlestat_trace_enable(false))
+ return -1;
+
+ /* At this point we should have some spurious wake up
+ * at the beginning of the traces and at the end (wake
+ * up all cpus and timer expiration for the timer
+ * acquisition). We assume these will be lost in the number
+ * of other traces and could be negligible. */
+ if (idlestat_store(options.filename))
+ return -1;
+ }
+
+ /* Load the idle states information */
+ datas = idlestat_load(options.filename);
+ if (!datas)
+ return 1;
+
+ /* Compute cluster idle intersection between cpus belonging to
+ * the same cluster
+ */
+ if (0 == establish_idledata_to_topo(datas)) {
+ if (options.dump > 0)
+ dump_cpu_topo_info(options.cstate, options.iterations,
+ dump_states);
+ else
+ dump_cpu_topo_info(options.cstate, options.iterations,
+ display_states);
+ } else {
+ cluster = cluster_data(datas);
+ if (!cluster)
+ return 1;
+
+ if (options.dump > 0) {
+ dump_all_data(datas, options.cstate,
+ options.iterations, dump_states);
+ dump_all_data(cluster, options.cstate,
+ options.iterations, dump_states);
+ } else {
+ dump_all_data(datas, options.cstate,
+ options.iterations, display_states);
+ dump_all_data(cluster, options.cstate,
+ options.iterations, display_states);
+ }
+
+ free(cluster->cstates);
+ free(cluster);
+ }
+
+ /* Computation could be heavy, let's give some information
+ * about the memory consumption */
+ if (options.debug) {
+ getrusage(RUSAGE_SELF, &rusage);
+ printf("max rss : %ld kB\n", rusage.ru_maxrss);
+ }
+
+ release_cpu_topo_cstates();
+ release_cpu_topo_info();
+ release_pstate_info(datas->pstates, datas->nrcpus);
+ release_cstate_info(datas->cstates, datas->nrcpus);
+ free(datas);
+
+ return 0;
+}
diff --git a/tools/power/idlestat/idlestat.h b/tools/power/idlestat/idlestat.h
new file mode 100644
index 0000000..58477a6
--- /dev/null
+++ b/tools/power/idlestat/idlestat.h
@@ -0,0 +1,106 @@
+/*
+ * idlestat.h
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#ifndef __IDLESTAT_H
+#define __IDLESTAT_H
+
+#define BUFSIZE 256
+#define NAMELEN 16
+#define MAXCSTATE 8
+#define MAXPSTATE 8
+#define MAX(A, B) (A > B ? A : B)
+#define MIN(A, B) (A < B ? A : B)
+#define AVG(A, B, I) ((A) + ((B - A) / (I)))
+
+#define IRQ_WAKEUP_UNIT_NAME "cpu"
+
+struct cpuidle_data {
+ double begin;
+ double end;
+ double duration;
+};
+
+struct cpuidle_cstate {
+ char *name;
+ struct cpuidle_data *data;
+ int nrdata;
+ double avg_time;
+ double max_time;
+ double min_time;
+ double duration;
+};
+
+enum IRQ_TYPE {
+ HARD_IRQ = 0,
+ IPI_IRQ,
+ IRQ_TYPE_MAX
+};
+
+struct wakeup_irq {
+ int id;
+ int irq_type;
+ char name[NAMELEN+1];
+ int count;
+};
+
+struct wakeup_info {
+ struct wakeup_irq *irqinfo;
+ int nrdata;
+};
+
+struct cpuidle_cstates {
+ struct cpuidle_cstate cstate[MAXCSTATE];
+ struct wakeup_info wakeinfo;
+ int last_cstate;
+ int cstate_max;
+ struct wakeup_irq *wakeirq;
+};
+
+struct cpufreq_pstate {
+ int id;
+ unsigned int freq;
+ int count;
+ double min_time;
+ double max_time;
+ double avg_time;
+ double duration;
+};
+
+struct cpufreq_pstates {
+ struct cpufreq_pstate *pstate;
+ int current;
+ int idle;
+ double time_enter;
+ double time_exit;
+ int max;
+};
+
+struct cpuidle_datas {
+ struct cpuidle_cstates *cstates;
+ struct cpufreq_pstates *pstates;
+ int nrcpus;
+};
+
+#endif
diff --git a/tools/power/idlestat/list.h b/tools/power/idlestat/list.h
new file mode 100644
index 0000000..45166da
--- /dev/null
+++ b/tools/power/idlestat/list.h
@@ -0,0 +1,588 @@
+/*
+ * list.h
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#ifndef _LINUX_LIST_H
+#define _LINUX_LIST_H
+
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+
+#define LIST_POISON1 ((void *)0x00100100)
+#define LIST_POISON2 ((void *)0x00200200)
+
+struct list_head {
+ struct list_head *next, *prev;
+};
+
+/*
+ * Simple doubly linked list implementation.
+ *
+ * Some of the internal functions ("__xxx") are useful when
+ * manipulating whole lists rather than single entries, as
+ * sometimes we already know the next/prev entries and we can
+ * generate better code by using them directly rather than
+ * using the generic single-entry routines.
+ */
+
+#define LIST_HEAD_INIT(name) { &(name), &(name) }
+
+#define LIST_HEAD(name) \
+ struct list_head name = LIST_HEAD_INIT(name)
+
+static inline void INIT_LIST_HEAD(struct list_head *list)
+{
+ list->next = list;
+ list->prev = list;
+}
+
+/*
+ * Insert a new entry between two known consecutive entries.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+static inline void __list_add(struct list_head *new,
+ struct list_head *prev,
+ struct list_head *next)
+{
+ next->prev = new;
+ new->next = next;
+ new->prev = prev;
+ prev->next = new;
+}
+
+/**
+ * list_add - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it after
+ *
+ * Insert a new entry after the specified head.
+ * This is good for implementing stacks.
+ */
+static inline void list_add(struct list_head *new, struct list_head *head)
+{
+ __list_add(new, head, head->next);
+}
+
+
+/**
+ * list_add_tail - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it before
+ *
+ * Insert a new entry before the specified head.
+ * This is useful for implementing queues.
+ */
+static inline void list_add_tail(struct list_head *new, struct list_head *head)
+{
+ __list_add(new, head->prev, head);
+}
+
+/*
+ * Delete a list entry by making the prev/next entries
+ * point to each other.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+static inline void __list_del(struct list_head *prev, struct list_head *next)
+{
+ next->prev = prev;
+ prev->next = next;
+}
+
+/**
+ * list_del - deletes entry from list.
+ * @entry: the element to delete from the list.
+ * Note: list_empty() on entry does not return true after this, the entry is
+ * in an undefined state.
+ */
+static inline void __list_del_entry(struct list_head *entry)
+{
+ __list_del(entry->prev, entry->next);
+}
+
+static inline void list_del(struct list_head *entry)
+{
+ __list_del(entry->prev, entry->next);
+ entry->next = LIST_POISON1;
+ entry->prev = LIST_POISON2;
+}
+
+/**
+ * list_replace - replace old entry by new one
+ * @old : the element to be replaced
+ * @new : the new element to insert
+ *
+ * If @old was empty, it will be overwritten.
+ */
+static inline void list_replace(struct list_head *old,
+ struct list_head *new)
+{
+ new->next = old->next;
+ new->next->prev = new;
+ new->prev = old->prev;
+ new->prev->next = new;
+}
+
+static inline void list_replace_init(struct list_head *old,
+ struct list_head *new)
+{
+ list_replace(old, new);
+ INIT_LIST_HEAD(old);
+}
+
+/**
+ * list_del_init - deletes entry from list and reinitialize it.
+ * @entry: the element to delete from the list.
+ */
+static inline void list_del_init(struct list_head *entry)
+{
+ __list_del_entry(entry);
+ INIT_LIST_HEAD(entry);
+}
+
+/**
+ * list_move - delete from one list and add as another's head
+ * @list: the entry to move
+ * @head: the head that will precede our entry
+ */
+static inline void list_move(struct list_head *list, struct list_head *head)
+{
+ __list_del_entry(list);
+ list_add(list, head);
+}
+
+/**
+ * list_move_tail - delete from one list and add as another's tail
+ * @list: the entry to move
+ * @head: the head that will follow our entry
+ */
+static inline void list_move_tail(struct list_head *list,
+ struct list_head *head)
+{
+ __list_del_entry(list);
+ list_add_tail(list, head);
+}
+
+/**
+ * list_is_last - tests whether @list is the last entry in list @head
+ * @list: the entry to test
+ * @head: the head of the list
+ */
+static inline int list_is_last(const struct list_head *list,
+ const struct list_head *head)
+{
+ return list->next == head;
+}
+
+/**
+ * list_empty - tests whether a list is empty
+ * @head: the list to test.
+ */
+static inline int list_empty(const struct list_head *head)
+{
+ return head->next == head;
+}
+
+/**
+ * list_empty_careful - tests whether a list is empty and not being modified
+ * @head: the list to test
+ *
+ * Description:
+ * tests whether a list is empty _and_ checks that no other CPU might be
+ * in the process of modifying either member (next or prev)
+ *
+ * NOTE: using list_empty_careful() without synchronization
+ * can only be safe if the only activity that can happen
+ * to the list entry is list_del_init(). Eg. it cannot be used
+ * if another CPU could re-list_add() it.
+ */
+static inline int list_empty_careful(const struct list_head *head)
+{
+ struct list_head *next = head->next;
+ return (next == head) && (next == head->prev);
+}
+
+/**
+ * list_rotate_left - rotate the list to the left
+ * @head: the head of the list
+ */
+static inline void list_rotate_left(struct list_head *head)
+{
+ struct list_head *first;
+
+ if (!list_empty(head)) {
+ first = head->next;
+ list_move_tail(first, head);
+ }
+}
+
+/**
+ * list_is_singular - tests whether a list has just one entry.
+ * @head: the list to test.
+ */
+static inline int list_is_singular(const struct list_head *head)
+{
+ return !list_empty(head) && (head->next == head->prev);
+}
+
+static inline void __list_cut_position(struct list_head *list,
+ struct list_head *head, struct list_head *entry)
+{
+ struct list_head *new_first = entry->next;
+ list->next = head->next;
+ list->next->prev = list;
+ list->prev = entry;
+ entry->next = list;
+ head->next = new_first;
+ new_first->prev = head;
+}
+
+/**
+ * list_cut_position - cut a list into two
+ * @list: a new list to add all removed entries
+ * @head: a list with entries
+ * @entry: an entry within head, could be the head itself
+ * and if so we won't cut the list
+ *
+ * This helper moves the initial part of @head, up to and
+ * including @entry, from @head to @list. You should
+ * pass on @entry an element you know is on @head. @list
+ * should be an empty list or a list you do not care about
+ * losing its data.
+ *
+ */
+static inline void list_cut_position(struct list_head *list,
+ struct list_head *head, struct list_head *entry)
+{
+ if (list_empty(head))
+ return;
+ if (list_is_singular(head) &&
+ (head->next != entry && head != entry))
+ return;
+ if (entry == head)
+ INIT_LIST_HEAD(list);
+ else
+ __list_cut_position(list, head, entry);
+}
+
+static inline void __list_splice(const struct list_head *list,
+ struct list_head *prev,
+ struct list_head *next)
+{
+ struct list_head *first = list->next;
+ struct list_head *last = list->prev;
+
+ first->prev = prev;
+ prev->next = first;
+
+ last->next = next;
+ next->prev = last;
+}
+
+/**
+ * list_splice - join two lists, this is designed for stacks
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ */
+static inline void list_splice(const struct list_head *list,
+ struct list_head *head)
+{
+ if (!list_empty(list))
+ __list_splice(list, head, head->next);
+}
+
+/**
+ * list_splice_tail - join two lists, each list being a queue
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ */
+static inline void list_splice_tail(struct list_head *list,
+ struct list_head *head)
+{
+ if (!list_empty(list))
+ __list_splice(list, head->prev, head);
+}
+
+/**
+ * list_splice_init - join two lists and reinitialise the emptied list.
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ *
+ * The list at @list is reinitialised
+ */
+static inline void list_splice_init(struct list_head *list,
+ struct list_head *head)
+{
+ if (!list_empty(list)) {
+ __list_splice(list, head, head->next);
+ INIT_LIST_HEAD(list);
+ }
+}
+
+/**
+ * list_splice_tail_init - join two lists and reinitialise the emptied list
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ *
+ * Each of the lists is a queue.
+ * The list at @list is reinitialised
+ */
+static inline void list_splice_tail_init(struct list_head *list,
+ struct list_head *head)
+{
+ if (!list_empty(list)) {
+ __list_splice(list, head->prev, head);
+ INIT_LIST_HEAD(list);
+ }
+}
+
+#undef offsetof
+#define offsetof(s, m) ((size_t)&(((s *)0)->m))
+
+#undef container_of
+#define container_of(ptr, type, member) ({ \
+ const typeof(((type *)0)->member) * __mptr = (ptr); \
+ (type *)((char *)__mptr - offsetof(type, member)); })
+
+/**
+ * list_entry - get the struct for this entry
+ * @ptr: the &struct list_head pointer.
+ * @type: the type of the struct this is embedded in.
+ * @member: the name of the list_struct within the struct.
+ */
+#define list_entry(ptr, type, member) \
+ container_of(ptr, type, member)
+
+/**
+ * list_first_entry - get the first element from a list
+ * @ptr: the list head to take the element from.
+ * @type: the type of the struct this is embedded in.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Note, that list is expected to be not empty.
+ */
+#define list_first_entry(ptr, type, member) \
+ list_entry((ptr)->next, type, member)
+
+/**
+ * list_for_each - iterate over a list
+ * @pos: the &struct list_head to use as a loop cursor.
+ * @head: the head for your list.
+ */
+#define list_for_each(pos, head) \
+ for (pos = (head)->next; pos != (head); pos = pos->next)
+
+/**
+ * __list_for_each - iterate over a list
+ * @pos: the &struct list_head to use as a loop cursor.
+ * @head: the head for your list.
+ *
+ * This variant doesn't differ from list_for_each() any more.
+ * We don't do prefetching in either case.
+ */
+#define __list_for_each(pos, head) \
+ for (pos = (head)->next; pos != (head); pos = pos->next)
+
+/**
+ * list_for_each_prev - iterate over a list backwards
+ * @pos: the &struct list_head to use as a loop cursor.
+ * @head: the head for your list.
+ */
+#define list_for_each_prev(pos, head) \
+ for (pos = (head)->prev; pos != (head); pos = pos->prev)
+
+/**
+ * list_for_each_safe - iterate over a list safe against removal of list entry
+ * @pos: the &struct list_head to use as a loop cursor.
+ * @n: another &struct list_head to use as temporary storage
+ * @head: the head for your list.
+ */
+#define list_for_each_safe(pos, n, head) \
+ for (pos = (head)->next, n = pos->next; pos != (head); \
+ pos = n, n = pos->next)
+
+/**
+ * list_for_each_prev_safe - iterate over a list backwards safe against removal of list entry
+ * @pos: the &struct list_head to use as a loop cursor.
+ * @n: another &struct list_head to use as temporary storage
+ * @head: the head for your list.
+ */
+#define list_for_each_prev_safe(pos, n, head) \
+ for (pos = (head)->prev, n = pos->prev; \
+ pos != (head); \
+ pos = n, n = pos->prev)
+
+/**
+ * list_for_each_entry - iterate over list of given type
+ * @pos: the type * to use as a loop cursor.
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ */
+#define list_for_each_entry(pos, head, member) \
+ for (pos = list_entry((head)->next, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = list_entry(pos->member.next, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_reverse - iterate backwards over list of given type.
+ * @pos: the type * to use as a loop cursor.
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ */
+#define list_for_each_entry_reverse(pos, head, member) \
+ for (pos = list_entry((head)->prev, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = list_entry(pos->member.prev, typeof(*pos), member))
+
+/**
+ * list_prepare_entry - prepare a entry for use in list_for_each_entry_continue()
+ * @pos: the type * to use as a start point
+ * @head: the head of the list
+ * @member: the name of the list_struct within the struct.
+ *
+ * Prepares a entry for use as a start point in list_for_each_entry_continue().
+ */
+#define list_prepare_entry(pos, head, member) \
+ ((pos) ? : list_entry(head, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_continue - continue iteration over list of given type
+ * @pos: the type * to use as a loop cursor.
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Continue to iterate over list of given type, continuing after
+ * the current position.
+ */
+#define list_for_each_entry_continue(pos, head, member) \
+ for (pos = list_entry(pos->member.next, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = list_entry(pos->member.next, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_continue_reverse - iterate backwards from the given point
+ * @pos: the type * to use as a loop cursor.
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Start to iterate over list of given type backwards, continuing after
+ * the current position.
+ */
+#define list_for_each_entry_continue_reverse(pos, head, member) \
+ for (pos = list_entry(pos->member.prev, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = list_entry(pos->member.prev, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_from - iterate over list of given type from the current point
+ * @pos: the type * to use as a loop cursor.
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Iterate over list of given type, continuing from current position.
+ */
+#define list_for_each_entry_from(pos, head, member) \
+ for (; &pos->member != (head); \
+ pos = list_entry(pos->member.next, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_safe - iterate over list of given type safe against removal of list entry
+ * @pos: the type * to use as a loop cursor.
+ * @n: another type * to use as temporary storage
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ */
+#define list_for_each_entry_safe(pos, n, head, member) \
+ for (pos = list_entry((head)->next, typeof(*pos), member), \
+ n = list_entry(pos->member.next, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = n, n = list_entry(n->member.next, typeof(*n), member))
+
+/**
+ * list_for_each_entry_safe_continue - continue list iteration safe against removal
+ * @pos: the type * to use as a loop cursor.
+ * @n: another type * to use as temporary storage
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Iterate over list of given type, continuing after current point,
+ * safe against removal of list entry.
+ */
+#define list_for_each_entry_safe_continue(pos, n, head, member) \
+ for (pos = list_entry(pos->member.next, typeof(*pos), member), \
+ n = list_entry(pos->member.next, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = n, n = list_entry(n->member.next, typeof(*n), member))
+
+/**
+ * list_for_each_entry_safe_from - iterate over list from current point safe against removal
+ * @pos: the type * to use as a loop cursor.
+ * @n: another type * to use as temporary storage
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Iterate over list of given type from current point, safe against
+ * removal of list entry.
+ */
+#define list_for_each_entry_safe_from(pos, n, head, member) \
+ for (n = list_entry(pos->member.next, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = n, n = list_entry(n->member.next, typeof(*n), member))
+
+/**
+ * list_for_each_entry_safe_reverse - iterate backwards over list safe against removal
+ * @pos: the type * to use as a loop cursor.
+ * @n: another type * to use as temporary storage
+ * @head: the head for your list.
+ * @member: the name of the list_struct within the struct.
+ *
+ * Iterate backwards over list of given type, safe against removal
+ * of list entry.
+ */
+#define list_for_each_entry_safe_reverse(pos, n, head, member) \
+ for (pos = list_entry((head)->prev, typeof(*pos), member), \
+ n = list_entry(pos->member.prev, typeof(*pos), member); \
+ &pos->member != (head); \
+ pos = n, n = list_entry(n->member.prev, typeof(*n), member))
+
+/**
+ * list_safe_reset_next - reset a stale list_for_each_entry_safe loop
+ * @pos: the loop cursor used in the list_for_each_entry_safe loop
+ * @n: temporary storage used in list_for_each_entry_safe
+ * @member: the name of the list_struct within the struct.
+ *
+ * list_safe_reset_next is not safe to use in general if the list may be
+ * modified concurrently (eg. the lock is dropped in the loop body). An
+ * exception to this is if the cursor element (pos) is pinned in the list,
+ * and list_safe_reset_next is called after re-taking the lock and before
+ * completing the current iteration of the loop body.
+ */
+#define list_safe_reset_next(pos, n, member) \
+ (n = list_entry(pos->member.next, typeof(*pos), member))
+
+#endif
diff --git a/tools/power/idlestat/topology.c b/tools/power/idlestat/topology.c
new file mode 100644
index 0000000..f610053
--- /dev/null
+++ b/tools/power/idlestat/topology.c
@@ -0,0 +1,503 @@
+/*
+ * topology.c
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#define _GNU_SOURCE
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <unistd.h>
+#include <string.h>
+#include <dirent.h>
+#include <ctype.h>
+#include <sys/stat.h>
+#include <assert.h>
+
+#include "list.h"
+#include "utils.h"
+#include "topology.h"
+#include "idlestat.h"
+
+struct cpu_topology g_cpu_topo_list;
+
+struct topology_info {
+ int physical_id;
+ int core_id;
+ int cpu_id;
+};
+
+struct list_info {
+ struct list_head hlist;
+ int id;
+};
+
+struct list_head *check_exist_from_head(struct list_head *head, int id)
+{
+ struct list_head *tmp;
+
+ list_for_each(tmp, head) {
+ if (id == ((struct list_info *)tmp)->id)
+ return tmp;
+ }
+
+ return NULL;
+}
+
+struct list_head *check_pos_from_head(struct list_head *head, int id)
+{
+ struct list_head *tmp;
+
+ list_for_each(tmp, head) {
+ if (id < ((struct list_info *)tmp)->id)
+ break;
+ }
+
+ return tmp->prev;
+}
+
+int add_topo_info(struct cpu_topology *topo_list, struct topology_info *info)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu = NULL;
+ struct list_head *ptr;
+
+ /* add cpu physical info */
+ ptr = check_exist_from_head(&topo_list->physical_head,
+ info->physical_id);
+ if (!ptr) {
+ s_phy = calloc(sizeof(struct cpu_physical), 1);
+ if (!s_phy)
+ return -1;
+
+ s_phy->core_num = 0;
+ s_phy->physical_id = info->physical_id;
+ INIT_LIST_HEAD(&s_phy->core_head);
+
+ ptr = check_pos_from_head(&topo_list->physical_head,
+ s_phy->physical_id);
+ list_add(&s_phy->list_physical, ptr);
+ topo_list->physical_num++;
+ } else {
+ s_phy = list_entry(ptr, struct cpu_physical,
+ list_physical);
+ }
+
+ /* add cpu core info */
+ ptr = check_exist_from_head(&s_phy->core_head, info->core_id);
+ if (!ptr) {
+ s_core = calloc(sizeof(struct cpu_core), 1);
+ if (!s_core)
+ return -1;
+
+ s_core->cpu_num = 0;
+ s_core->is_ht = false;
+ s_core->core_id = info->core_id;
+ INIT_LIST_HEAD(&s_core->cpu_head);
+
+ ptr = check_pos_from_head(&s_phy->core_head,
+ s_core->core_id);
+ list_add(&s_core->list_core, ptr);
+ s_phy->core_num++;
+
+ } else {
+ s_core = list_entry(ptr, struct cpu_core, list_core);
+ }
+
+ /* add cpu info */
+ ptr = check_exist_from_head(&s_core->cpu_head, info->cpu_id);
+ if (!ptr) {
+ s_cpu = calloc(sizeof(struct cpu_cpu), 1);
+ if (!s_cpu)
+ return -1;
+
+ s_cpu->cpu_id = info->cpu_id;
+
+ ptr = check_pos_from_head(&s_core->cpu_head, s_cpu->cpu_id);
+ list_add(&s_cpu->list_cpu, ptr);
+ s_core->cpu_num++;
+ if (s_core->cpu_num > 1)
+ s_core->is_ht = true;
+ }
+
+ return 0;
+}
+
+void free_cpu_cpu_list(struct list_head *head)
+{
+ struct cpu_cpu *lcpu, *n;
+
+ list_for_each_entry_safe(lcpu, n, head, list_cpu) {
+ list_del(&lcpu->list_cpu);
+ free(lcpu);
+ }
+}
+
+void free_cpu_core_list(struct list_head *head)
+{
+ struct cpu_core *lcore, *n;
+
+ list_for_each_entry_safe(lcore, n, head, list_core) {
+ free_cpu_cpu_list(&lcore->cpu_head);
+ list_del(&lcore->list_core);
+ free(lcore);
+ }
+}
+
+void free_cpu_topology(struct list_head *head)
+{
+ struct cpu_physical *lphysical, *n;
+
+ list_for_each_entry_safe(lphysical, n, head, list_physical) {
+ free_cpu_core_list(&lphysical->core_head);
+ list_del(&lphysical->list_physical);
+ free(lphysical);
+ }
+}
+
+int output_topo_info(struct cpu_topology *topo_list)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu;
+
+ list_for_each_entry(s_phy, &topo_list->physical_head, list_physical) {
+ printf("cluster%c:\n", s_phy->physical_id + 'A');
+ list_for_each_entry(s_core, &s_phy->core_head, list_core) {
+ printf("\tcore%d\n", s_core->core_id);
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu)
+ printf("\t\tcpu%d\n", s_cpu->cpu_id);
+ }
+ }
+
+ return 0;
+}
+
+int outfile_topo_info(FILE *f, struct cpu_topology *topo_list)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu;
+
+ list_for_each_entry(s_phy, &topo_list->physical_head, list_physical) {
+ fprintf(f, "cluster%c:\n", s_phy->physical_id + 'A');
+ list_for_each_entry(s_core, &s_phy->core_head, list_core) {
+ fprintf(f, "\tcore%d\n", s_core->core_id);
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu)
+ fprintf(f, "\t\tcpu%d\n", s_cpu->cpu_id);
+ }
+ }
+
+ return 0;
+}
+
+struct cpu_cpu *find_cpu_point(struct cpu_topology *topo_list, int cpuid)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu;
+
+ list_for_each_entry(s_phy, &topo_list->physical_head, list_physical)
+ list_for_each_entry(s_core, &s_phy->core_head, list_core)
+ list_for_each_entry(s_cpu, &s_core->cpu_head, list_cpu)
+ if (s_cpu->cpu_id == cpuid)
+ return s_cpu;
+
+ return NULL;
+}
+
+static inline int read_topology_cb(char *path, struct topology_info *info)
+{
+ file_read_value(path, "core_id", "%d", &info->core_id);
+ file_read_value(path, "physical_package_id", "%d", &info->physical_id);
+
+ return 0;
+}
+
+typedef int (*folder_filter_t)(const char *name);
+
+static int cpu_filter_cb(const char *name)
+{
+ /* let's ignore some directories in order to avoid to be
+ * pulled inside the sysfs circular symlinks mess/hell
+ * (choose the word which fit better)*/
+ if (!strcmp(name, "cpuidle"))
+ return 1;
+
+ if (!strcmp(name, "cpufreq"))
+ return 1;
+
+ return 0;
+}
+
+/*
+ * This function will browse the directory structure and build a
+ * reflecting the content of the directory tree.
+ *
+ * @path : the root node of the folder
+ * @filter : a callback to filter out the directories
+ * Returns 0 on success, -1 otherwise
+ */
+static int topo_folder_scan(char *path, folder_filter_t filter)
+{
+ DIR *dir, *dir_topology;
+ char *basedir, *newpath;
+ struct dirent dirent, *direntp;
+ struct stat s;
+ int ret = 0;
+
+ dir = opendir(path);
+ if (!dir) {
+ printf("error: unable to open directory %s\n", path);
+ return -1;
+ }
+
+ ret = asprintf(&basedir, "%s", path);
+ if (ret < 0)
+ return -1;
+
+ while (!readdir_r(dir, &dirent, &direntp)) {
+
+ if (!direntp)
+ break;
+
+ if (direntp->d_name[0] == '.')
+ continue;
+
+ if (filter && filter(direntp->d_name))
+ continue;
+
+ if (!strstr(direntp->d_name, "cpu"))
+ continue;
+
+ ret = asprintf(&newpath, "%s/%s/%s", basedir,
+ direntp->d_name, "topology");
+ if (ret < 0)
+ goto out_free_basedir;
+
+ ret = stat(newpath, &s);
+ if (ret)
+ goto out_free_newpath;
+
+ if (S_ISDIR(s.st_mode) || (S_ISLNK(s.st_mode))) {
+ struct topology_info cpu_info;
+
+ dir_topology = opendir(path);
+ if (!dir_topology)
+ continue;
+
+ read_topology_cb(newpath, &cpu_info);
+ assert(sscanf(direntp->d_name, "cpu%d",
+ &cpu_info.cpu_id) == 1);
+ add_topo_info(&g_cpu_topo_list, &cpu_info);
+ }
+
+out_free_newpath:
+ free(newpath);
+
+ if (ret)
+ break;
+ }
+
+out_free_basedir:
+ free(basedir);
+
+ closedir(dir);
+
+ return ret;
+}
+
+
+int init_cpu_topo_info(void)
+{
+ INIT_LIST_HEAD(&g_cpu_topo_list.physical_head);
+ g_cpu_topo_list.physical_num = 0;
+
+ return 0;
+}
+
+int read_sysfs_cpu_topo(void)
+{
+ topo_folder_scan("/sys/devices/system/cpu", cpu_filter_cb);
+
+ return 0;
+}
+
+int read_cpu_topo_info(FILE *f, char *buf)
+{
+ int ret = 0;
+ struct topology_info cpu_info;
+ bool is_ht = false;
+ char pid;
+
+ do {
+ ret = sscanf(buf, "cluster%c", &pid);
+ if (!ret)
+ break;
+
+ cpu_info.physical_id = pid - 'A';
+
+ fgets(buf, BUFSIZE, f);
+ do {
+ ret = sscanf(buf, "\tcore%u", &cpu_info.core_id);
+ if (ret) {
+ is_ht = true;
+ fgets(buf, BUFSIZE, f);
+ } else {
+ ret = sscanf(buf, "\tcpu%u", &cpu_info.cpu_id);
+ if (ret)
+ is_ht = false;
+ else
+ break;
+ }
+
+ do {
+ if (!is_ht) {
+ ret = sscanf(buf, "\tcpu%u",
+ &cpu_info.cpu_id);
+ cpu_info.core_id = cpu_info.cpu_id;
+ } else {
+ ret = sscanf(buf, "\t\tcpu%u",
+ &cpu_info.cpu_id);
+ }
+
+ if (!ret)
+ break;
+
+ add_topo_info(&g_cpu_topo_list, &cpu_info);
+
+ fgets(buf, BUFSIZE, f);
+ } while (1);
+ } while (1);
+ } while (1);
+
+ /* output_topo_info(&g_cpu_topo_list); */
+
+ return 0;
+}
+
+int release_cpu_topo_info(void)
+{
+ /* free alloced memory */
+ free_cpu_topology(&g_cpu_topo_list.physical_head);
+
+ return 0;
+}
+
+int output_cpu_topo_info(FILE *f)
+{
+ outfile_topo_info(f, &g_cpu_topo_list);
+
+ return 0;
+}
+
+int establish_idledata_to_topo(struct cpuidle_datas *datas)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu;
+ int i;
+ int has_topo = 0;
+
+ for (i = 0; i < datas->nrcpus; i++) {
+ s_cpu = find_cpu_point(&g_cpu_topo_list, i);
+ if (s_cpu) {
+ s_cpu->cstates = &datas->cstates[i];
+ s_cpu->pstates = &datas->pstates[i];
+ has_topo = 1;
+ }
+ }
+
+ if (!has_topo)
+ return -1;
+
+ list_for_each_entry(s_phy, &g_cpu_topo_list.physical_head,
+ list_physical)
+ list_for_each_entry(s_core, &s_phy->core_head, list_core)
+ s_core->cstates = core_cluster_data(s_core);
+
+ list_for_each_entry(s_phy, &g_cpu_topo_list.physical_head,
+ list_physical)
+ s_phy->cstates = physical_cluster_data(s_phy);
+
+ return 0;
+}
+
+int dump_cpu_topo_info(int state, int count,
+ int (*dump)(struct cpuidle_cstates *, struct cpufreq_pstates *,
+ int, int, char *))
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+ struct cpu_cpu *s_cpu;
+ char tmp[30];
+ int tab = 0;
+
+ list_for_each_entry(s_phy, &g_cpu_topo_list.physical_head,
+ list_physical) {
+ sprintf(tmp, "cluster%c", s_phy->physical_id + 'A');
+ dump(s_phy->cstates, NULL, state, count, tmp);
+
+ list_for_each_entry(s_core, &s_phy->core_head, list_core) {
+ if (s_core->is_ht) {
+ sprintf(tmp, " core%d", s_core->core_id);
+ dump(s_core->cstates, NULL, state, count, tmp);
+
+ tab = 1;
+ } else {
+ tab = 0;
+ }
+
+ list_for_each_entry(s_cpu, &s_core->cpu_head,
+ list_cpu) {
+ sprintf(tmp, "%*ccpu%d", (tab + 1) * 2, 0x20,
+ s_cpu->cpu_id);
+ dump(s_cpu->cstates, s_cpu->pstates, state,
+ count, tmp);
+ }
+ }
+ }
+
+ return 0;
+}
+
+int release_cpu_topo_cstates(void)
+{
+ struct cpu_physical *s_phy;
+ struct cpu_core *s_core;
+
+ list_for_each_entry(s_phy, &g_cpu_topo_list.physical_head,
+ list_physical) {
+ free(s_phy->cstates);
+ s_phy->cstates = NULL;
+ list_for_each_entry(s_core, &s_phy->core_head, list_core)
+ if (s_core->is_ht) {
+ free(s_core->cstates);
+ s_core->cstates = NULL;
+ }
+ }
+
+ return 0;
+}
diff --git a/tools/power/idlestat/topology.h b/tools/power/idlestat/topology.h
new file mode 100644
index 0000000..5ad244c
--- /dev/null
+++ b/tools/power/idlestat/topology.h
@@ -0,0 +1,77 @@
+/*
+ * topology.h
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#ifndef __TOPOLOGY_H
+#define __TOPOLOGY_H
+
+#include "list.h"
+#include "idlestat.h"
+
+struct cpu_cpu {
+ struct list_head list_cpu;
+ int cpu_id;
+ struct cpuidle_cstates *cstates;
+ struct cpufreq_pstates *pstates;
+};
+
+struct cpu_core {
+ struct list_head list_core;
+ int core_id;
+ struct list_head cpu_head;
+ int cpu_num;
+ bool is_ht;
+ struct cpuidle_cstates *cstates;
+};
+
+struct cpu_physical {
+ struct list_head list_physical;
+ int physical_id;
+ struct list_head core_head;
+ int core_num;
+ struct cpuidle_cstates *cstates;
+};
+
+struct cpu_topology {
+ struct list_head physical_head;
+ int physical_num;
+};
+
+extern int init_cpu_topo_info(void);
+extern int read_cpu_topo_info(FILE *f, char *buf);
+extern int read_sysfs_cpu_topo(void);
+extern int release_cpu_topo_info(void);
+extern int output_cpu_topo_info(FILE *f);
+extern int establish_idledata_to_topo(struct cpuidle_datas *datas);
+extern int release_cpu_topo_cstates(void);
+extern int dump_cpu_topo_info(int state, int count,
+ int (*dump)(struct cpuidle_cstates *, struct cpufreq_pstates *,
+ int, int, char *));
+
+
+extern struct cpuidle_cstates *core_cluster_data(struct cpu_core *s_core);
+extern struct cpuidle_cstates *
+ physical_cluster_data(struct cpu_physical *s_phy);
+
+#endif
diff --git a/tools/power/idlestat/trace.c b/tools/power/idlestat/trace.c
new file mode 100644
index 0000000..741415f
--- /dev/null
+++ b/tools/power/idlestat/trace.c
@@ -0,0 +1,87 @@
+/*
+ * trace.c
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <unistd.h>
+#include <string.h>
+
+#include "trace.h"
+#include "utils.h"
+
+int idlestat_trace_enable(bool enable)
+{
+ return write_int(TRACE_ON_PATH, enable);
+}
+
+int idlestat_flush_trace(void)
+{
+ return write_int(TRACE_FILE, 0);
+}
+
+int idlestat_init_trace(unsigned int duration)
+{
+ int bufsize;
+
+ /* Assuming the worst case where we can have for cpuidle,
+ * TRACE_IDLE_NRHITS_PER_SEC. Each state enter/exit line are
+ * 196 chars wide, so we have 2 x 196 x TRACE_IDLE_NRHITS_PER_SEC lines.
+ * For cpufreq, assume a 196-character line for each frequency change,
+ * and expect a rate of TRACE_CPUFREQ_NRHITS_PER_SEC.
+ * Divide by 2^10 to have Kb. We add 1Kb to be sure to round up.
+ */
+
+ bufsize = 2 * TRACE_IDLE_LENGTH * TRACE_IDLE_NRHITS_PER_SEC;
+ bufsize += TRACE_CPUFREQ_LENGTH * TRACE_CPUFREQ_NRHITS_PER_SEC;
+ bufsize = (bufsize * duration / (1 << 10)) + 1;
+
+ if (write_int(TRACE_BUFFER_SIZE_PATH, bufsize))
+ return -1;
+
+ if (read_int(TRACE_BUFFER_TOTAL_PATH, &bufsize))
+ return -1;
+
+ printf("Total trace buffer: %d kB\n", bufsize);
+
+ /* Disable all the traces */
+ if (write_int(TRACE_EVENT_PATH, 0))
+ return -1;
+
+ /* Enable cpu_idle traces */
+ if (write_int(TRACE_CPUIDLE_EVENT_PATH, 1))
+ return -1;
+
+ /* Enable cpu_frequency traces */
+ if (write_int(TRACE_CPUFREQ_EVENT_PATH, 1))
+ return -1;
+
+ /* Enable irq traces */
+ if (write_int(TRACE_IRQ_EVENT_PATH, 1))
+ return -1;
+
+ return 0;
+}
diff --git a/tools/power/idlestat/trace.h b/tools/power/idlestat/trace.h
new file mode 100644
index 0000000..aa805347
--- /dev/null
+++ b/tools/power/idlestat/trace.h
@@ -0,0 +1,43 @@
+/*
+ * trace.h
+ *
+ * Copyright (C) 2014 Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+#ifndef __TRACE_H
+#define __TRACE_H
+
+#define TRACE_PATH "/sys/kernel/debug/tracing"
+#define TRACE_ON_PATH TRACE_PATH "/tracing_on"
+#define TRACE_BUFFER_SIZE_PATH TRACE_PATH "/buffer_size_kb"
+#define TRACE_BUFFER_TOTAL_PATH TRACE_PATH "/buffer_total_size_kb"
+#define TRACE_CPUIDLE_EVENT_PATH TRACE_PATH "/events/power/cpu_idle/enable"
+#define TRACE_CPUFREQ_EVENT_PATH TRACE_PATH "/events/power/cpu_frequency/enable"
+#define TRACE_IRQ_EVENT_PATH TRACE_PATH "/events/irq/enable"
+#define TRACE_EVENT_PATH TRACE_PATH "/events/enable"
+#define TRACE_FREE TRACE_PATH "/free_buffer"
+#define TRACE_FILE TRACE_PATH "/trace"
+#define TRACE_IDLE_NRHITS_PER_SEC 10000
+#define TRACE_IDLE_LENGTH 196
+#define TRACE_CPUFREQ_NRHITS_PER_SEC 100
+#define TRACE_CPUFREQ_LENGTH 196
+
+extern int idlestat_trace_enable(bool enable);
+extern int idlestat_flush_trace(void);
+extern int idlestat_init_trace(unsigned int duration);
+
+#endif
diff --git a/tools/power/idlestat/utils.c b/tools/power/idlestat/utils.c
new file mode 100644
index 0000000..a82afe6
--- /dev/null
+++ b/tools/power/idlestat/utils.c
@@ -0,0 +1,115 @@
+/*
+ * utils.c
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#define _GNU_SOURCE
+#include <stdio.h>
+#undef _GNU_SOURCE
+#include <stdlib.h>
+
+#include "utils.h"
+
+int write_int(const char *path, int val)
+{
+ FILE *f;
+
+ f = fopen(path, "w");
+ if (!f) {
+ fprintf(stderr, "failed to open '%s': %m\n", path);
+ return -1;
+ }
+
+ fprintf(f, "%d", val);
+
+ fclose(f);
+
+ return 0;
+}
+
+int read_int(const char *path, int *val)
+{
+ FILE *f;
+
+ f = fopen(path, "r");
+
+ if (!f) {
+ fprintf(stderr, "failed to open '%s': %m\n", path);
+ return -1;
+ }
+
+ fscanf(f, "%d", val);
+
+ fclose(f);
+
+ return 0;
+}
+
+int store_line(const char *line, void *data)
+{
+ FILE *f = data;
+
+ /* ignore comment line */
+ if (line[0] == '#')
+ return 0;
+
+ fprintf(f, "%s", line);
+
+ return 0;
+}
+
+/*
+ * This functions is a helper to read a specific file content and store
+ * the content inside a variable pointer passed as parameter, the format
+ * parameter gives the variable type to be read from the file.
+ *
+ * @path : directory path containing the file
+ * @name : name of the file to be read
+ * @format : the format of the format
+ * @value : a pointer to a variable to store the content of the file
+ * Returns 0 on success, -1 otherwise
+ */
+int file_read_value(const char *path, const char *name,
+ const char *format, void *value)
+{
+ FILE *file;
+ char *rpath;
+ int ret;
+
+ ret = asprintf(&rpath, "%s/%s", path, name);
+ if (ret < 0)
+ return ret;
+
+ file = fopen(rpath, "r");
+ if (!file) {
+ ret = -1;
+ goto out_free;
+ }
+
+ ret = fscanf(file, format, value) == EOF ? -1 : 0;
+
+ fclose(file);
+out_free:
+ free(rpath);
+ return ret;
+}
diff --git a/tools/power/idlestat/utils.h b/tools/power/idlestat/utils.h
new file mode 100644
index 0000000..81a25fe
--- /dev/null
+++ b/tools/power/idlestat/utils.h
@@ -0,0 +1,35 @@
+/*
+ * utils.h
+ *
+ * Copyright (C) 2014, Linaro Limited.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ * 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; version 2 of the License.
+ *
+ * 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.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * Contributors:
+ * Daniel Lezcano <daniel.lezcano@linaro.org>
+ * Zoran Markovic <zoran.markovic@linaro.org>
+ *
+ */
+#ifndef __UTILS_H
+#define __UTILS_H
+
+extern int write_int(const char *path, int val);
+extern int read_int(const char *path, int *val);
+extern int store_line(const char *line, void *data);
+extern int file_read_value(const char *path, const char *name,
+ const char *format, void *value);
+
+#endif
--
1.7.9.5
^ permalink raw reply related [flat|nested] 4+ messages in thread
* [RFC PATCH 2/2] sched: Add documentation for idlestat scheduler benchmarking tool
2014-03-24 20:05 [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 1/2] power: Add idlestat tool for benchmarking energy-aware scheduler Zoran Markovic
@ 2014-03-24 20:05 ` Zoran Markovic
2014-03-25 12:08 ` [RFC PATCH 0/2] sched: proposal " Preeti Murthy
2 siblings, 0 replies; 4+ messages in thread
From: Zoran Markovic @ 2014-03-24 20:05 UTC (permalink / raw)
To: linux-kernel; +Cc: rob, mingo, peterz, rostedt, daniel.lezcano, zoran.markovic
This patch documents the proposed functionality of idlestat tool and
states its intended use for scheduler benchmarking. The documentation
file describes the design of the tool, what kernel functionality it
relies upon, and what information is contained in the output report.
It also contains a simple linear model for estimating CPU power
consumption during idlestat run.
Idlestat focuses itself on CPU and cluster power states in precise
intervals in time. This is of particular use when the benchmarked
process is a load synthesis tool: idlestat could focus its acquisition
period to a particular sub-period in the load sequence. Output results
from idlestat can be applied to a power model in order to estimate the
power consumption of CPUs and clusters during the benchmark interval.
Initial measurements on ARM Versatile Express TC2 platform show a model
error of ~2.6% for the linear power model described in the documentation.
Cc: Rob Landley <rob@landley.net>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Zoran Markovic <zoran.markovic@linaro.org>
---
Documentation/scheduler/idlestat.txt | 79 ++++++++++++++++++++++++++++++++++
1 file changed, 79 insertions(+)
create mode 100644 Documentation/scheduler/idlestat.txt
diff --git a/Documentation/scheduler/idlestat.txt b/Documentation/scheduler/idlestat.txt
new file mode 100644
index 0000000..8e6b695
--- /dev/null
+++ b/Documentation/scheduler/idlestat.txt
@@ -0,0 +1,79 @@
+This document captures the desired operation of the idlestat tool.
+
+With the advent of battery-powered Linux devices, it became important to add
+a power-aware component to the existing CFS scheduler solution. Future
+developments in this field need to be benchmarked using a simple tool that
+monitors power parameters during system runs and provides sufficient info for
+developers to assess how changes to scheduler code affected CPU power
+consumption. The idlestat tool attempts to capture this.
+
+Idlestat uses kernel's FTRACE function to monitor and capture C-state and
+P-state transitions of CPUs over a time interval. It extracts the following
+information from trace file:
+ - Times when CPUs entered and exited a certain C-state
+ - Times when CPUs entered and exited a certain P-state
+ - Raised IRQs
+
+Following a successful run, idlestat calculates and reports the following
+information:
+ - Total, average, minimum and maximum time spent in each C-state,
+ per-CPU.
+ - Total, average, minimum and maximum time spent in each P-state,
+ per-CPU.
+ - Total, average, minimum and maximum time during which all CPUs in
+ a cluster were in the same C-state, per-cluster.
+ - Number of times a certain IRQ caused a CPU to exit idle state,
+ per-CPU and per-IRQ.
+
+The tool parses sysfs entries to determine the CPU/cluster topology, as well
+as supported C-states and P-states per CPU. It is unaware of CPU/cluster power
+consumption in each C-state and P-state, but if these parameters are
+externally known, a ballpark estimate of the energy consumed during idlestat
+run can be calculated as follows:
+
+energy = sum_per_cpu(PCi*(TCi-TCCi)) + sum_per_cluster(PCCi*TCCi) +
+ sum_per_cpu(PPi*TPi)
+
+where:
+PCi - is the power consumption of CPU in Ci power state
+TCi - is the total time the CPU has spent in Ci power state
+PCCi - is the power consumption of cluster in Ci power state
+TCCi - is the total time the cluster has spent in Ci power state
+PPi - is the power consumption of CPU in Pi power state
+TPi - is the total time the CPU has spent in Pi power state
+
+Below is an example report of one idlestat run on a dual-core system:
+clusterA@state hits total(us) avg(us) min(us) max(us)
+ C1 10821 5879554.00 543.35 0.00 23163.00
+ C2 0 0.00 0.00 0.00 0.00
+ C3 78 2929290.00 37555.00 0.00 101441.00
+ cpu0@state hits total(us) avg(us) min(us) max(us)
+ C1 6744 6407808.00 950.15 0.00 23194.00
+ C2 3 8819.00 2939.67 549.00 5310.00
+ C3 75 2960110.00 39468.13 213.00 101441.00
+ 350 1047 204490.00 195.31 0.00 4578.00
+ 700 5628 396247.00 70.41 0.00 1465.00
+ 920 0 0.00 0.00 0.00 0.00
+ cpu0 wakeups name count
+ irq109 ehci_hcd:usb1 1727
+ irq029 twd 4524
+ irq069 gp_timer 60
+ irq115 mmc0 7
+ irq044 DMA 3
+ cpu1@state hits total(us) avg(us) min(us) max(us)
+ C1 6544 6398931.00 977.83 0.00 36255.00
+ C2 1 1129.00 1129.00 1129.00 1129.00
+ C3 77 2955293.00 38380.43 122.00 101471.00
+ 350 1124 212428.00 188.99 0.00 18677.00
+ 700 5366 408782.00 76.18 0.00 946.00
+ 920 0 0.00 0.00 0.00 0.00
+ cpu1 wakeups name count
+ irq029 twd 4737
+
+Idlestat does not perform any processing during the acquisition period. It
+sleeps while traces are captured, making sure it is non-intrusive to C-
+and P-state transitions. During that time, traces are stored in kernel ring
+buffers previously sized by idlestat based on the length of acquisition
+period and estimated frequency of trace events. Traces are parsed and
+analyzed once the acquisition period is complete.
+
--
1.7.9.5
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool
2014-03-24 20:05 [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 1/2] power: Add idlestat tool for benchmarking energy-aware scheduler Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 2/2] sched: Add documentation for idlestat scheduler benchmarking tool Zoran Markovic
@ 2014-03-25 12:08 ` Preeti Murthy
2 siblings, 0 replies; 4+ messages in thread
From: Preeti Murthy @ 2014-03-25 12:08 UTC (permalink / raw)
To: Zoran Markovic
Cc: LKML, Rob Landley, mingo, Peter Zijlstra, rostedt,
Daniel Lezcano, Preeti U Murthy
Hi Zoran,
I understand that this approach is non-intrusive with the
running workload.
However it would be nice to know how my system is behaving
in terms of power efficiency at a given instance of time and
accordingly I can take action to kill a few applications to save
power, as against knowing how my system did a a minute ago
for example.
For this I would require a tool which shows me say
every 5seconds how my system is doing. Just like top or
turbostat, which have a time interval for polling and show
results over the polling period. This would enable the
user to take immediate action along with seeing the result
of it from the same tool.
Would it not be possible to parse the /proc stats and the
/sys/devices/cpu/cpu*/cpufreq_or_cpuidle stats to get the
information about the idle and pstates just like you do to
get the topology information but continuously through
periodic polling ? Would this be too intrusive considering
top and such commonly used tools do it this way?
You could also perhaps say that the intention of this
is to verify the correctness of power aware algorithms
like the power aware scheduler in which case this approach
of collecting the trace after a given duration will be of use.
Having said that, a tool that gives the running power efficiency
image of my system would be more useful in the long run.
Regards
Preeti U Murthy
On Tue, Mar 25, 2014 at 1:35 AM, Zoran Markovic
<zoran.markovic@linaro.org> wrote:
> Conclusions from Energy Aware Scheduling sessions at the latest Kernel Summit
> identified a need for tools that would assess power consumption of the system
> These tools would be used to prove efficiency of scheduler patches by
> comparing power consumption before and after they were applied.
>
> Attached is the proposal for the idlestat tool. The purpose of this patch
> is to solicit feedback on tool's features, possible enhancements, etc.
>
> Source code and sample idlestat report are provided for reference.
>
> Please review and provide comments in anticipation of further development.
>
> Regards, Zoran
>
> Zoran Markovic (2):
> power: Add idlestat tool for benchmarking energy-aware scheduler
> sched: Add documentation for idlestat scheduler benchmarking tool
>
> Documentation/scheduler/idlestat.txt | 79 +++
> tools/power/idlestat/.gitignore | 50 ++
> tools/power/idlestat/Makefile | 34 +
> tools/power/idlestat/idlestat.c | 1229 ++++++++++++++++++++++++++++++++++
> tools/power/idlestat/idlestat.h | 106 +++
> tools/power/idlestat/list.h | 588 ++++++++++++++++
> tools/power/idlestat/topology.c | 503 ++++++++++++++
> tools/power/idlestat/topology.h | 77 +++
> tools/power/idlestat/trace.c | 87 +++
> tools/power/idlestat/trace.h | 43 ++
> tools/power/idlestat/utils.c | 115 ++++
> tools/power/idlestat/utils.h | 35 +
> 12 files changed, 2946 insertions(+)
> create mode 100644 Documentation/scheduler/idlestat.txt
> create mode 100644 tools/power/idlestat/.gitignore
> create mode 100644 tools/power/idlestat/Makefile
> create mode 100644 tools/power/idlestat/idlestat.c
> create mode 100644 tools/power/idlestat/idlestat.h
> create mode 100644 tools/power/idlestat/list.h
> create mode 100644 tools/power/idlestat/topology.c
> create mode 100644 tools/power/idlestat/topology.h
> create mode 100644 tools/power/idlestat/trace.c
> create mode 100644 tools/power/idlestat/trace.h
> create mode 100644 tools/power/idlestat/utils.c
> create mode 100644 tools/power/idlestat/utils.h
>
> --
> 1.7.9.5
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2014-03-25 12:08 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-03-24 20:05 [RFC PATCH 0/2] sched: proposal for idlestat scheduler benchmarking tool Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 1/2] power: Add idlestat tool for benchmarking energy-aware scheduler Zoran Markovic
2014-03-24 20:05 ` [RFC PATCH 2/2] sched: Add documentation for idlestat scheduler benchmarking tool Zoran Markovic
2014-03-25 12:08 ` [RFC PATCH 0/2] sched: proposal " Preeti Murthy
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).