Merge git://git.kernel.org/pub/scm/linux/kernel/git/brodo/cpupowerutils

* git://git.kernel.org/pub/scm/linux/kernel/git/brodo/cpupowerutils:
  cpupower: Do detect IDA (opportunistic processor performance) via cpuid
  cpupower: Show Intel turbo ratio support via ./cpupower frequency-info
  cpupowerutils: increase MAX_LINE_LEN
  cpupower: Rename package from cpupowerutils to cpupower
  cpupowerutils: Rename: libcpufreq->libcpupower
  cpupowerutils: use kernel version-derived version string
  cpupowerutils: utils - ConfigStyle bugfixes
  cpupowerutils: helpers - ConfigStyle bugfixes
  cpupowerutils: idle_monitor - ConfigStyle bugfixes
  cpupowerutils: lib - ConfigStyle bugfixes
  cpupowerutils: bench - ConfigStyle bugfixes
  cpupowerutils: do not update po files on each and every compile
  cpupowerutils: remove ccdv, use kernel quiet/verbose mechanism
  cpupowerutils: use COPYING, CREDITS from top-level directory
  cpupowerutils - cpufrequtils extended with quite some features

CREDITS

@@ -504,7 +504,7 @@ N: Dominik Brodowski
 E: linux@brodo.de
 W: http://www.brodo.de/
 P: 1024D/725B37C6  190F 3E77 9C89 3B6D BECD  46EE 67C3 0308 725B 37C6
-D: parts of CPUFreq code, ACPI bugfixes
+D: parts of CPUFreq code, ACPI bugfixes, PCMCIA rewrite, cpufrequtils
 S: Tuebingen, Germany
 
 N: Andries Brouwer
@@ -857,6 +857,10 @@ S: One Dell Way
 S: Round Rock, TX  78682
 S: USA
 
+N: Mattia Dongili
+E: malattia@gmail.com
+D: cpufrequtils (precursor to cpupowerutils)
+
 N: Ben Dooks
 E: ben-linux@fluff.org
 E: ben@simtec.co.uk
@@ -1883,6 +1887,11 @@ S: Kruislaan 419
 S: 1098 VA Amsterdam 
 S: The Netherlands
 
+N: Goran Koruga
+E: korugag@siol.net
+D: cpufrequtils (precursor to cpupowerutils)
+S: Slovenia
+
 N: Jiri Kosina
 E: jikos@jikos.cz
 E: jkosina@suse.cz
@@ -2916,6 +2925,12 @@ S: Schlossbergring 9
 S: 79098 Freiburg
 S: Germany
 
+N: Thomas Renninger
+E: trenn@suse.de
+D: cpupowerutils
+S: SUSE Linux GmbH
+S: Germany
+
 N: Joerg Reuter
 E: jreuter@yaina.de
 W: http://yaina.de/jreuter/

MAINTAINERS

@@ -1925,6 +1925,12 @@ S:	Maintained
 F:	arch/x86/kernel/cpuid.c
 F:	arch/x86/kernel/msr.c
 
+CPU POWER MONITORING SUBSYSTEM
+M:	Dominik Brodowski <linux@dominikbrodowski.net>
+M:	Thomas Renninger <trenn@suse.de>
+S:	Maintained
+F:	tools/power/cpupower
+
 CPUSETS
 M:	Paul Menage <menage@google.com>
 W:	http://www.bullopensource.org/cpuset/

tools/power/cpupower/.gitignore

+.libs
+libcpupower.so
+libcpupower.so.0
+libcpupower.so.0.0.0
+build/ccdv
+cpufreq-info
+cpufreq-set
+cpufreq-aperf
+lib/.libs
+lib/cpufreq.lo
+lib/cpufreq.o
+lib/proc.lo
+lib/proc.o
+lib/sysfs.lo
+lib/sysfs.o
+po/cpupowerutils.pot
+po/*.gmo
+utils/cpufreq-info.o
+utils/cpufreq-set.o
+utils/cpufreq-aperf.o
+cpupower
+bench/cpufreq-bench

tools/power/cpupower/Makefile

+# Makefile for cpupower
+#
+# Copyright (C) 2005,2006 Dominik Brodowski <linux@dominikbrodowski.net>
+#
+# Based largely on the Makefile for udev by:
+#
+# Copyright (C) 2003,2004 Greg Kroah-Hartman <greg@kroah.com>
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; 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; if not, write to the Free Software
+# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+#
+
+# --- CONFIGURATION BEGIN ---
+
+# Set the following to `true' to make a unstripped, unoptimized
+# binary. Leave this set to `false' for production use.
+DEBUG ?=	false
+
+# make the build silent. Set this to something else to make it noisy again.
+V ?=		false
+
+# Internationalization support (output in different languages).
+# Requires gettext.
+NLS ?=		true
+
+# Set the following to 'true' to build/install the
+# cpufreq-bench benchmarking tool
+CPUFRQ_BENCH ?= true
+
+# Prefix to the directories we're installing to
+DESTDIR ?=
+
+# --- CONFIGURATION END ---
+
+
+
+# Package-related definitions. Distributions can modify the version
+# and _should_ modify the PACKAGE_BUGREPORT definition
+
+VERSION=			$(shell ./utils/version-gen.sh)
+LIB_MAJ=			0.0.0
+LIB_MIN=			0
+
+PACKAGE =			cpupower
+PACKAGE_BUGREPORT =		cpufreq@vger.kernel.org
+LANGUAGES = 			de fr it cs pt
+
+
+# Directory definitions. These are default and most probably
+# do not need to be changed. Please note that DESTDIR is
+# added in front of any of them
+
+bindir ?=	/usr/bin
+sbindir ?=	/usr/sbin
+mandir ?=	/usr/man
+includedir ?=	/usr/include
+libdir ?=	/usr/lib
+localedir ?=	/usr/share/locale
+docdir ?=       /usr/share/doc/packages/cpupower
+confdir ?=      /etc/
+
+# Toolchain: what tools do we use, and what options do they need:
+
+CP = cp -fpR
+INSTALL = /usr/bin/install -c
+INSTALL_PROGRAM = ${INSTALL}
+INSTALL_DATA  = ${INSTALL} -m 644
+INSTALL_SCRIPT = ${INSTALL_PROGRAM}
+
+# If you are running a cross compiler, you may want to set this
+# to something more interesting, like "arm-linux-".  If you want
+# to compile vs uClibc, that can be done here as well.
+CROSS = #/usr/i386-linux-uclibc/usr/bin/i386-uclibc-
+CC = $(CROSS)gcc
+LD = $(CROSS)gcc
+AR = $(CROSS)ar
+STRIP = $(CROSS)strip
+RANLIB = $(CROSS)ranlib
+HOSTCC = gcc
+
+
+# Now we set up the build system
+#
+
+# set up PWD so that older versions of make will work with our build.
+PWD = $(shell pwd)
+
+GMO_FILES = ${shell for HLANG in ${LANGUAGES}; do echo po/$$HLANG.gmo; done;}
+
+export CROSS CC AR STRIP RANLIB CFLAGS LDFLAGS LIB_OBJS
+
+# check if compiler option is supported
+cc-supports = ${shell if $(CC) ${1} -S -o /dev/null -xc /dev/null > /dev/null 2>&1; then echo "$(1)"; fi;}
+
+# use '-Os' optimization if available, else use -O2
+OPTIMIZATION := $(call cc-supports,-Os,-O2)
+
+WARNINGS := -Wall -Wchar-subscripts -Wpointer-arith -Wsign-compare
+WARNINGS += $(call cc-supports,-Wno-pointer-sign)
+WARNINGS += $(call cc-supports,-Wdeclaration-after-statement)
+WARNINGS += -Wshadow
+
+CFLAGS += -DVERSION=\"$(VERSION)\" -DPACKAGE=\"$(PACKAGE)\" \
+		-DPACKAGE_BUGREPORT=\"$(PACKAGE_BUGREPORT)\" -D_GNU_SOURCE
+
+UTIL_OBJS =  utils/helpers/amd.o utils/helpers/topology.o utils/helpers/msr.o \
+	utils/helpers/sysfs.o utils/helpers/misc.o utils/helpers/cpuid.o \
+	utils/helpers/pci.o utils/helpers/bitmask.o \
+	utils/idle_monitor/nhm_idle.o utils/idle_monitor/snb_idle.o \
+	utils/idle_monitor/amd_fam14h_idle.o utils/idle_monitor/cpuidle_sysfs.o \
+	utils/idle_monitor/mperf_monitor.o utils/idle_monitor/cpupower-monitor.o \
+	utils/cpupower.o utils/cpufreq-info.o utils/cpufreq-set.o \
+	utils/cpupower-set.o utils/cpupower-info.o utils/cpuidle-info.o
+
+UTIL_HEADERS = utils/helpers/helpers.h utils/idle_monitor/cpupower-monitor.h \
+	utils/helpers/bitmask.h \
+	utils/idle_monitor/idle_monitors.h utils/idle_monitor/idle_monitors.def
+
+UTIL_SRC := $(UTIL_OBJS:.o=.c)
+
+LIB_HEADERS = 	lib/cpufreq.h lib/sysfs.h
+LIB_SRC = 	lib/cpufreq.c lib/sysfs.c
+LIB_OBJS = 	lib/cpufreq.o lib/sysfs.o
+
+CFLAGS +=	-pipe
+
+ifeq ($(strip $(NLS)),true)
+	INSTALL_NLS += install-gmo
+	COMPILE_NLS += create-gmo
+endif
+
+ifeq ($(strip $(CPUFRQ_BENCH)),true)
+	INSTALL_BENCH += install-bench
+	COMPILE_BENCH += compile-bench
+endif
+
+CFLAGS += $(WARNINGS)
+
+ifeq ($(strip $(V)),false)
+	QUIET=@
+	ECHO=@echo
+else
+	QUIET=
+	ECHO=@\#
+endif
+export QUIET ECHO
+
+# if DEBUG is enabled, then we do not strip or optimize
+ifeq ($(strip $(DEBUG)),true)
+	CFLAGS += -O1 -g -DDEBUG
+	STRIPCMD = /bin/true -Since_we_are_debugging
+else
+	CFLAGS += $(OPTIMIZATION) -fomit-frame-pointer
+	STRIPCMD = $(STRIP) -s --remove-section=.note --remove-section=.comment
+endif
+
+
+# the actual make rules
+
+all: libcpupower cpupower $(COMPILE_NLS) $(COMPILE_BENCH)
+
+lib/%.o: $(LIB_SRC) $(LIB_HEADERS)
+	$(ECHO) "  CC      " $@
+	$(QUIET) $(CC) $(CFLAGS) -fPIC -o $@ -c lib/$*.c
+
+libcpupower.so.$(LIB_MAJ): $(LIB_OBJS)
+	$(ECHO) "  LD      " $@
+	$(QUIET) $(CC) -shared $(CFLAGS) $(LDFLAGS) -o $@ \
+		-Wl,-soname,libcpupower.so.$(LIB_MIN) $(LIB_OBJS)
+	@ln -sf $@ libcpupower.so
+	@ln -sf $@ libcpupower.so.$(LIB_MIN)
+
+libcpupower: libcpupower.so.$(LIB_MAJ)
+
+# Let all .o files depend on its .c file and all headers
+# Might be worth to put this into utils/Makefile at some point of time
+$(UTIL_OBJS): $(UTIL_HEADERS)
+
+.c.o:
+	$(ECHO) "  CC      " $@
+	$(QUIET) $(CC) $(CFLAGS) -I./lib -I ./utils -o $@ -c $*.c
+
+cpupower: $(UTIL_OBJS) libcpupower.so.$(LIB_MAJ)
+	$(ECHO) "  CC      " $@
+	$(QUIET) $(CC) $(CFLAGS) $(LDFLAGS) -lcpupower -lrt -lpci -L. -o $@ $(UTIL_OBJS)
+	$(QUIET) $(STRIPCMD) $@
+
+po/$(PACKAGE).pot: $(UTIL_SRC)
+	$(ECHO) "  GETTEXT " $@
+	$(QUIET) xgettext --default-domain=$(PACKAGE) --add-comments \
+		--keyword=_ --keyword=N_ $(UTIL_SRC) && \
+	test -f $(PACKAGE).po && \
+	mv -f $(PACKAGE).po po/$(PACKAGE).pot
+
+po/%.gmo: po/%.po
+	$(ECHO) "  MSGFMT  " $@
+	$(QUIET) msgfmt -o $@ po/$*.po
+
+create-gmo: ${GMO_FILES}
+
+update-po: po/$(PACKAGE).pot
+	$(ECHO) "  MSGMRG  " $@
+	$(QUIET) @for HLANG in $(LANGUAGES); do \
+		echo -n "Updating $$HLANG "; \
+		if msgmerge po/$$HLANG.po po/$(PACKAGE).pot -o \
+		   po/$$HLANG.new.po; then \
+			mv -f po/$$HLANG.new.po po/$$HLANG.po; \
+		else \
+			echo "msgmerge for $$HLANG failed!"; \
+			rm -f po/$$HLANG.new.po; \
+		fi; \
+	done;
+
+compile-bench: libcpupower.so.$(LIB_MAJ)
+	@V=$(V) confdir=$(confdir) $(MAKE) -C bench
+
+clean:
+	-find . \( -not -type d \) -and \( -name '*~' -o -name '*.[oas]' \) -type f -print \
+	 | xargs rm -f
+	-rm -f $(UTIL_BINS)
+	-rm -f $(IDLE_OBJS)
+	-rm -f cpupower
+	-rm -f libcpupower.so*
+	-rm -rf po/*.gmo po/*.pot
+	$(MAKE) -C bench clean
+
+
+install-lib:
+	$(INSTALL) -d $(DESTDIR)${libdir}
+	$(CP) libcpupower.so* $(DESTDIR)${libdir}/
+	$(INSTALL) -d $(DESTDIR)${includedir}
+	$(INSTALL_DATA) lib/cpufreq.h $(DESTDIR)${includedir}/cpufreq.h
+
+install-tools:
+	$(INSTALL) -d $(DESTDIR)${bindir}
+	$(INSTALL_PROGRAM) cpupower $(DESTDIR)${bindir}
+
+install-man:
+	$(INSTALL_DATA) -D man/cpupower.1 $(DESTDIR)${mandir}/man1/cpupower.1
+	$(INSTALL_DATA) -D man/cpupower-frequency-set.1 $(DESTDIR)${mandir}/man1/cpupower-frequency-set.1
+	$(INSTALL_DATA) -D man/cpupower-frequency-info.1 $(DESTDIR)${mandir}/man1/cpupower-frequency-info.1
+	$(INSTALL_DATA) -D man/cpupower-set.1 $(DESTDIR)${mandir}/man1/cpupower-set.1
+	$(INSTALL_DATA) -D man/cpupower-info.1 $(DESTDIR)${mandir}/man1/cpupower-info.1
+	$(INSTALL_DATA) -D man/cpupower-monitor.1 $(DESTDIR)${mandir}/man1/cpupower-monitor.1
+
+install-gmo:
+	$(INSTALL) -d $(DESTDIR)${localedir}
+	for HLANG in $(LANGUAGES); do \
+		echo '$(INSTALL_DATA) -D po/$$HLANG.gmo $(DESTDIR)${localedir}/$$HLANG/LC_MESSAGES/cpupower.mo'; \
+		$(INSTALL_DATA) -D po/$$HLANG.gmo $(DESTDIR)${localedir}/$$HLANG/LC_MESSAGES/cpupower.mo; \
+	done;
+
+install-bench:
+	@#DESTDIR must be set from outside to survive
+	@sbindir=$(sbindir) bindir=$(bindir) docdir=$(docdir) confdir=$(confdir) $(MAKE) -C bench install
+
+install: all install-lib install-tools install-man $(INSTALL_NLS) $(INSTALL_BENCH)
+
+uninstall:
+	- rm -f $(DESTDIR)${libdir}/libcpupower.*
+	- rm -f $(DESTDIR)${includedir}/cpufreq.h
+	- rm -f $(DESTDIR)${bindir}/utils/cpupower
+	- rm -f $(DESTDIR)${mandir}/man1/cpufreq-set.1
+	- rm -f $(DESTDIR)${mandir}/man1/cpufreq-info.1
+	- for HLANG in $(LANGUAGES); do \
+		rm -f $(DESTDIR)${localedir}/$$HLANG/LC_MESSAGES/cpupower.mo; \
+	  done;
+
+.PHONY: all utils libcpupower update-po create-gmo install-lib install-tools install-man install-gmo install uninstall clean

tools/power/cpupower/README

+The cpufrequtils package (homepage: 
+http://www.kernel.org/pub/linux/utils/kernel/cpufreq/cpufrequtils.html ) 
+consists of the following elements:
+
+requirements
+------------
+
+On x86 pciutils is needed at runtime (-lpci).
+For compilation pciutils-devel (pci/pci.h) and a gcc version
+providing cpuid.h is needed.
+For both it's not explicitly checked for (yet).
+
+
+libcpufreq
+----------
+
+"libcpufreq" is a library which offers a unified access method for userspace
+tools and programs to the cpufreq core and drivers in the Linux kernel. This
+allows for code reduction in userspace tools, a clean implementation of
+the interaction to the cpufreq core, and support for both the sysfs and proc
+interfaces [depending on configuration, see below].
+
+
+compilation and installation
+----------------------------
+
+make
+su
+make install
+
+should suffice on most systems. It builds default libcpufreq,
+cpufreq-set and cpufreq-info files and installs them in /usr/lib and
+/usr/bin, respectively. If you want to set up the paths differently and/or
+want to configure the package to your specific needs, you need to open
+"Makefile" with an editor of your choice and edit the block marked
+CONFIGURATION.
+
+
+THANKS
+------
+Many thanks to Mattia Dongili who wrote the autotoolization and
+libtoolization, the manpages and the italian language file for cpufrequtils;
+to Dave Jones for his feedback and his dump_psb tool; to Bruno Ducrot for his
+powernow-k8-decode and intel_gsic tools as well as the french language file;
+and to various others commenting on the previous (pre-)releases of 
+cpufrequtils.
+
+
+        Dominik Brodowski

tools/power/cpupower/ToDo

+ToDos sorted by priority:
+
+- Use bitmask functions to parse CPU topology more robust
+  (current implementation has issues on AMD)
+- Try to read out boost states and frequencies on Intel
+- Adjust README
+- Somewhere saw the ability to read power consumption of
+  RAM from HW on Intel SandyBridge -> another monitor?
+- Add another c1e debug idle monitor
+  -> Is by design racy with BIOS, but could be added
+     with a --force option and some "be careful" messages

tools/power/cpupower/bench/Makefile

+LIBS = -L../ -lm -lcpupower
+
+OBJS = main.o parse.o system.o benchmark.o
+CFLAGS += -D_GNU_SOURCE -I../lib -DDEFAULT_CONFIG_FILE=\"$(confdir)/cpufreq-bench.conf\"
+
+%.o : %.c
+	$(ECHO) "  CC      " $@
+	$(QUIET) $(CC) -c $(CFLAGS) $< -o $@
+
+cpufreq-bench: $(OBJS)
+	$(ECHO) "  CC      " $@
+	$(QUIET) $(CC) -o $@ $(CFLAGS) $(OBJS) $(LIBS)
+
+all: cpufreq-bench
+
+install:
+	mkdir -p $(DESTDIR)/$(sbindir)
+	mkdir -p $(DESTDIR)/$(bindir)
+	mkdir -p $(DESTDIR)/$(docdir)
+	mkdir -p $(DESTDIR)/$(confdir)
+	install -m 755 cpufreq-bench $(DESTDIR)/$(sbindir)/cpufreq-bench
+	install -m 755 cpufreq-bench_plot.sh $(DESTDIR)/$(bindir)/cpufreq-bench_plot.sh
+	install -m 644 README-BENCH $(DESTDIR)/$(docdir)/README-BENCH
+	install -m 755 cpufreq-bench_script.sh $(DESTDIR)/$(docdir)/cpufreq-bench_script.sh
+	install -m 644 example.cfg $(DESTDIR)/$(confdir)/cpufreq-bench.conf
+
+clean:
+	rm -f *.o
+	rm -f cpufreq-bench

tools/power/cpupower/bench/README-BENCH

+This is cpufreq-bench, a microbenchmark for the cpufreq framework.
+
+Purpose
+=======
+
+What is this benchmark for:
+  - Identify worst case performance loss when doing dynamic frequency
+    scaling using Linux kernel governors
+  - Identify average reaction time of a governor to CPU load changes
+  - (Stress) Testing whether a cpufreq low level driver or governor works
+    as expected
+  - Identify cpufreq related performance regressions between kernels
+  - Possibly Real time priority testing? -> what happens if there are
+    processes with a higher prio than the governor's kernel thread
+  - ...
+
+What this benchmark does *not* cover:
+  - Power saving related regressions (In fact as better the performance
+    throughput is, the worse the power savings will be, but the first should
+    mostly count more...)
+  - Real world (workloads)
+
+
+Description
+===========
+
+cpufreq-bench helps to test the condition of a given cpufreq governor.
+For that purpose, it compares the performance governor to a configured
+powersave module.
+
+
+How it works
+============
+You can specify load (100% CPU load) and sleep (0% CPU load) times in us which
+will be run X time in a row (cycles):
+
+         sleep=25000
+         load=25000
+         cycles=20
+
+This part of the configuration file will create 25ms load/sleep turns,
+repeated 20 times.
+
+Adding this:
+         sleep_step=25000
+         load_step=25000
+         rounds=5
+Will increase load and sleep time by 25ms 5 times.
+Together you get following test:
+25ms  load/sleep time repeated 20 times (cycles).
+50ms  load/sleep time repeated 20 times (cycles).
+..
+100ms load/sleep time repeated 20 times (cycles).
+
+First it is calibrated how long a specific CPU intensive calculation
+takes on this machine and needs to be run in a loop using the performance
+governor.
+Then the above test runs are processed using the performance governor
+and the governor to test. The time the calculation really needed
+with the dynamic freq scaling governor is compared with the time needed
+on full performance and you get the overall performance loss.
+
+
+Example of expected results with ondemand governor:
+
+This shows expected results of the first two test run rounds from
+above config, you there have:
+
+100% CPU load (load) | 0 % CPU load (sleep)  | round
+   25 ms             |    25 ms              |   1
+   50 ms             |    50 ms              |   2
+
+For example if ondemand governor is configured to have a 50ms
+sampling rate you get:
+
+In round 1, ondemand should have rather static 50% load and probably
+won't ever switch up (as long as up_threshold is above).
+
+In round 2, if the ondemand sampling times exactly match the load/sleep
+trigger of the cpufreq-bench, you will see no performance loss (compare with
+below possible ondemand sample kick ins (1)):
+
+But if ondemand always kicks in in the middle of the load sleep cycles, it
+will always see 50% loads and you get worst performance impact never
+switching up (compare with below possible ondemand sample kick ins (2))::
+
+      50     50   50   50ms ->time
+load -----|     |-----|     |-----|     |-----|
+          |     |     |     |     |     |     |
+sleep     |-----|     |-----|     |-----|     |----
+    |-----|-----|-----|-----|-----|-----|-----|----  ondemand sampling (1)
+         100    0    100    0    100    0    100     load seen by ondemand(%)
+       |-----|-----|-----|-----|-----|-----|-----|--   ondemand sampling (2)
+      50     50    50    50    50    50    50        load seen by ondemand(%)
+
+You can easily test all kind of load/sleep times and check whether your
+governor in average behaves as expected.
+
+
+ToDo
+====
+
+Provide a gnuplot utility script for easy generation of plots to present
+the outcome nicely.
+
+
+cpufreq-bench Command Usage
+===========================
+-l, --load=<long int>           initial load time in us
+-s, --sleep=<long int>          initial sleep time in us
+-x, --load-step=<long int>      time to be added to load time, in us
+-y, --sleep-step=<long int>     time to be added to sleep time, in us
+-c, --cpu=<unsigned int>        CPU Number to use, starting at 0
+-p, --prio=<priority>           scheduler priority, HIGH, LOW or DEFAULT
+-g, --governor=<governor>       cpufreq governor to test
+-n, --cycles=<int>              load/sleep cycles to get an avarage value to compare
+-r, --rounds<int>               load/sleep rounds
+-f, --file=<configfile>         config file to use
+-o, --output=<dir>              output dir, must exist
+-v, --verbose                   verbose output on/off
+
+Due to the high priority, the application may not be responsible for some time.
+After the benchmark, the logfile is saved in OUTPUTDIR/benchmark_TIMESTAMP.log
+

tools/power/cpupower/bench/benchmark.c

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <math.h>
+
+#include "config.h"
+#include "system.h"
+#include "benchmark.h"
+
+/* Print out progress if we log into a file */
+#define show_progress(total_time, progress_time)	\
+if (config->output != stdout) {				\
+	fprintf(stdout, "Progress: %02lu %%\r",		\
+		(progress_time * 100) / total_time);	\
+	fflush(stdout);					\
+}
+
+/**
+ * compute how many rounds of calculation we should do
+ * to get the given load time
+ *
+ * @param load aimed load time in µs
+ *
+ * @retval rounds of calculation
+ **/
+
+unsigned int calculate_timespace(long load, struct config *config)
+{
+	int i;
+	long long now, then;
+	unsigned int estimated = GAUGECOUNT;
+	unsigned int rounds = 0;
+	unsigned int timed = 0;
+
+	if (config->verbose)
+		printf("calibrating load of %lius, please wait...\n", load);
+
+	/* get the initial calculation time for a specific number of rounds */
+	now = get_time();
+	ROUNDS(estimated);
+	then = get_time();
+
+	timed = (unsigned int)(then - now);
+
+	/* approximation of the wanted load time by comparing with the
+	 * initial calculation time */
+	for (i = 0; i < 4; i++) {
+		rounds = (unsigned int)(load * estimated / timed);
+		dprintf("calibrating with %u rounds\n", rounds);
+		now = get_time();
+		ROUNDS(rounds);
+		then = get_time();
+
+		timed = (unsigned int)(then - now);
+		estimated = rounds;
+	}
+	if (config->verbose)
+		printf("calibration done\n");
+
+	return estimated;
+}
+
+/**
+ * benchmark
+ * generates a specific sleep an load time with the performance
+ * governor and compares the used time for same calculations done
+ * with the configured powersave governor
+ *
+ * @param config config values for the benchmark
+ *
+ **/
+
+void start_benchmark(struct config *config)
+{
+	unsigned int _round, cycle;
+	long long now, then;
+	long sleep_time = 0, load_time = 0;
+	long performance_time = 0, powersave_time = 0;
+	unsigned int calculations;
+	unsigned long total_time = 0, progress_time = 0;
+
+	sleep_time = config->sleep;
+	load_time = config->load;
+
+	/* For the progress bar */
+	for (_round = 1; _round <= config->rounds; _round++)
+		total_time += _round * (config->sleep + config->load);
+	total_time *= 2; /* powersave and performance cycles */
+
+	for (_round = 0; _round < config->rounds; _round++) {
+		performance_time = 0LL;
+		powersave_time = 0LL;
+
+		show_progress(total_time, progress_time);
+
+		/* set the cpufreq governor to "performance" which disables
+		 * P-State switching. */
+		if (set_cpufreq_governor("performance", config->cpu) != 0)
+			return;
+
+		/* calibrate the calculation time. the resulting calculation
+		 * _rounds should produce a load which matches the configured
+		 * load time */
+		calculations = calculate_timespace(load_time, config);
+
+		if (config->verbose)
+			printf("_round %i: doing %u cycles with %u calculations"
+			       " for %lius\n", _round + 1, config->cycles,
+			       calculations, load_time);
+
+		fprintf(config->output, "%u %li %li ",
+			_round, load_time, sleep_time);
+
+		if (config->verbose)
+			printf("avarage: %lius, rps:%li\n",
+				load_time / calculations,
+				1000000 * calculations / load_time);
+
+		/* do some sleep/load cycles with the performance governor */
+		for (cycle = 0; cycle < config->cycles; cycle++) {
+			now = get_time();
+			usleep(sleep_time);
+			ROUNDS(calculations);
+			then = get_time();
+			performance_time += then - now - sleep_time;
+			if (config->verbose)
+				printf("performance cycle took %lius, "
+					"sleep: %lius, "
+					"load: %lius, rounds: %u\n",
+					(long)(then - now), sleep_time,
+					load_time, calculations);
+		}
+		fprintf(config->output, "%li ",
+			performance_time / config->cycles);
+
+		progress_time += sleep_time + load_time;
+		show_progress(total_time, progress_time);
+
+		/* set the powersave governor which activates P-State switching
+		 * again */
+		if (set_cpufreq_governor(config->governor, config->cpu) != 0)
+			return;
+
+		/* again, do some sleep/load cycles with the
+		 * powersave governor */
+		for (cycle = 0; cycle < config->cycles; cycle++) {
+			now = get_time();
+			usleep(sleep_time);
+			ROUNDS(calculations);
+			then = get_time();
+			powersave_time += then - now - sleep_time;
+			if (config->verbose)
+				printf("powersave cycle took %lius, "
+					"sleep: %lius, "
+					"load: %lius, rounds: %u\n",
+					(long)(then - now), sleep_time,
+					load_time, calculations);
+		}
+
+		progress_time += sleep_time + load_time;
+
+		/* compare the avarage sleep/load cycles  */
+		fprintf(config->output, "%li ",
+			powersave_time / config->cycles);
+		fprintf(config->output, "%.3f\n",
+			performance_time * 100.0 / powersave_time);
+		fflush(config->output);
+
+		if (config->verbose)
+			printf("performance is at %.2f%%\n",
+				performance_time * 100.0 / powersave_time);
+
+		sleep_time += config->sleep_step;
+		load_time += config->load_step;
+	}
+}

tools/power/cpupower/bench/benchmark.h

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+/* load loop, this schould take about 1 to 2ms to complete */
+#define ROUNDS(x) {unsigned int rcnt;			       \
+		for (rcnt = 0; rcnt < x*1000; rcnt++) { \
+			(void)(((int)(pow(rcnt, rcnt) * \
+				      sqrt(rcnt*7230970)) ^ 7230716) ^ \
+				      (int)atan2(rcnt, rcnt));	       \
+		} }							\
+
+
+void start_benchmark(struct config *config);

tools/power/cpupower/bench/config.h

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+/* initial loop count for the load calibration */
+#define GAUGECOUNT	1500
+
+/* default scheduling policy SCHED_OTHER */
+#define SCHEDULER	SCHED_OTHER
+
+#define PRIORITY_DEFAULT 0
+#define PRIORITY_HIGH	 sched_get_priority_max(SCHEDULER)
+#define PRIORITY_LOW	 sched_get_priority_min(SCHEDULER)
+
+/* enable further debug messages */
+#ifdef DEBUG
+#define dprintf printf
+#else
+#define dprintf(...) do { } while (0)
+#endif
+

tools/power/cpupower/bench/cpufreq-bench_plot.sh

+#!/bin/bash
+
+# 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, or (at your option)
+# any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+# 02110-1301, USA.
+
+# Author/Copyright(c): 2009, Thomas Renninger <trenn@suse.de>, Novell Inc.
+
+# Helper script to easily create nice plots of your cpufreq-bench results
+
+dir=`mktemp -d`
+output_file="cpufreq-bench.png"
+global_title="cpufreq-bench plot"
+picture_type="jpeg"
+file[0]=""
+
+function usage()
+{
+    echo "cpufreq-bench_plot.sh [OPTIONS] logfile [measure_title] [logfile [measure_title]] ...]"
+    echo
+    echo "Options"
+    echo "   -o output_file"
+    echo "   -t global_title"
+    echo "   -p picture_type [jpeg|gif|png|postscript|...]"
+    exit 1
+}
+
+if [ $# -eq 0 ];then
+	echo "No benchmark results file provided"
+	echo
+	usage
+fi
+
+while getopts o:t:p: name ; do
+    case $name in
+	o)
+	    output_file="$OPTARG".$picture_type
+	    ;;
+	t)
+	    global_title="$OPTARG"
+	    ;;
+	p)
+	    picture_type="$OPTARG"
+	    ;;
+        ?)
+	    usage
+	    ;;
+    esac
+done
+shift $(($OPTIND -1))
+
+plots=0
+while [ "$1" ];do
+    if [ ! -f "$1" ];then
+	echo "File $1 does not exist"
+	usage
+    fi
+    file[$plots]="$1"
+    title[$plots]="$2"
+    # echo "File: ${file[$plots]} - ${title[plots]}"
+    shift;shift
+    plots=$((plots + 1))
+done
+
+echo "set terminal $picture_type"	>> $dir/plot_script.gpl
+echo "set output \"$output_file\""	>> $dir/plot_script.gpl
+echo "set title \"$global_title\""	>> $dir/plot_script.gpl
+echo "set xlabel \"sleep/load time\""	>> $dir/plot_script.gpl
+echo "set ylabel \"Performance (%)\""	>> $dir/plot_script.gpl
+
+for((plot=0;plot<$plots;plot++));do
+
+    # Sanity check
+    ###### I am to dump to get this redirected to stderr/stdout in one awk call... #####
+    cat ${file[$plot]} |grep -v "^#" |awk '{if ($2 != $3) printf("Error in measure %d:Load time %s does not equal sleep time %s, plot will not be correct\n", $1, $2, $3); ERR=1}'
+    ###### I am to dump to get this redirected in one awk call... #####
+
+    # Parse out load time (which must be equal to sleep time for a plot), divide it by 1000
+    # to get ms and parse out the performance in percentage and write it to a temp file for plotting
+    cat ${file[$plot]} |grep -v "^#" |awk '{printf "%lu %.1f\n",$2/1000, $6}' >$dir/data_$plot
+
+    if [ $plot -eq 0 ];then
+	echo -n "plot " >> $dir/plot_script.gpl
+    fi
+    echo -n "\"$dir/data_$plot\" title \"${title[$plot]}\" with lines" >> $dir/plot_script.gpl
+    if [ $(($plot + 1)) -ne $plots ];then
+	echo -n ", " >> $dir/plot_script.gpl
+    fi
+done
+echo >> $dir/plot_script.gpl
+
+gnuplot $dir/plot_script.gpl
+rm -r $dir
\ No newline at end of file

tools/power/cpupower/bench/cpufreq-bench_script.sh

+#!/bin/bash
+
+# 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, or (at your option)
+# any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+# 02110-1301, USA.
+
+# Author/Copyright(c): 2009, Thomas Renninger <trenn@suse.de>, Novell Inc.
+
+# Ondemand up_threshold and sampling rate test script for cpufreq-bench
+# mircobenchmark.
+# Modify the general variables at the top or extend or copy out parts
+# if you want to test other things
+#
+
+# Default with latest kernels is 95, before micro account patches
+# it was 80, cmp. with git commit 808009131046b62ac434dbc796
+UP_THRESHOLD="60 80 95"
+# Depending on the kernel and the HW sampling rate could be restricted
+# and cannot be set that low...
+# E.g. before git commit cef9615a853ebc4972084f7 one could only set
+# min sampling rate of 80000 if CONFIG_HZ=250
+SAMPLING_RATE="20000 80000"
+
+function measure()
+{
+    local -i up_threshold_set
+    local -i sampling_rate_set
+
+    for up_threshold in $UP_THRESHOLD;do
+	for sampling_rate in $SAMPLING_RATE;do
+	    # Set values in sysfs
+	    echo $up_threshold >/sys/devices/system/cpu/cpu0/cpufreq/ondemand/up_threshold
+	    echo $sampling_rate >/sys/devices/system/cpu/cpu0/cpufreq/ondemand/sampling_rate
+	    up_threshold_set=$(cat /sys/devices/system/cpu/cpu0/cpufreq/ondemand/up_threshold)
+	    sampling_rate_set=$(cat /sys/devices/system/cpu/cpu0/cpufreq/ondemand/sampling_rate)
+
+	    # Verify set values in sysfs
+	    if [ ${up_threshold_set} -eq ${up_threshold} ];then
+		echo "up_threshold: $up_threshold, set in sysfs: ${up_threshold_set}"
+	    else
+		echo "WARNING: Tried to set up_threshold: $up_threshold, set in sysfs: ${up_threshold_set}"
+	    fi
+	    if [ ${sampling_rate_set} -eq ${sampling_rate} ];then
+		echo "sampling_rate: $sampling_rate, set in sysfs: ${sampling_rate_set}"
+	    else
+		echo "WARNING: Tried to set sampling_rate: $sampling_rate, set in sysfs: ${sampling_rate_set}"
+	    fi
+
+	    # Benchmark
+	    cpufreq-bench -o /var/log/cpufreq-bench/up_threshold_${up_threshold}_sampling_rate_${sampling_rate}
+	done
+    done
+}
+
+function create_plots()
+{
+    local command
+
+    for up_threshold in $UP_THRESHOLD;do
+	command="cpufreq-bench_plot.sh -o \"sampling_rate_${SAMPLING_RATE}_up_threshold_${up_threshold}\" -t \"Ondemand sampling_rate: ${SAMPLING_RATE} comparison - Up_threshold: $up_threshold %\""
+	for sampling_rate in $SAMPLING_RATE;do
+	    command="${command} /var/log/cpufreq-bench/up_threshold_${up_threshold}_sampling_rate_${sampling_rate}/* \"sampling_rate = $sampling_rate\""
+	done
+	echo $command
+	eval "$command"
+	echo
+    done
+
+    for sampling_rate in $SAMPLING_RATE;do
+	command="cpufreq-bench_plot.sh -o \"up_threshold_${UP_THRESHOLD}_sampling_rate_${sampling_rate}\" -t \"Ondemand up_threshold: ${UP_THRESHOLD} % comparison - sampling_rate: $sampling_rate\""
+	for up_threshold in $UP_THRESHOLD;do
+	    command="${command} /var/log/cpufreq-bench/up_threshold_${up_threshold}_sampling_rate_${sampling_rate}/* \"up_threshold = $up_threshold\""
+	done
+	echo $command
+	eval "$command"
+	echo
+    done
+
+    command="cpufreq-bench_plot.sh -o \"up_threshold_${UP_THRESHOLD}_sampling_rate_${SAMPLING_RATE}\" -t \"Ondemand up_threshold: ${UP_THRESHOLD} and sampling_rate ${SAMPLING_RATE} comparison\""
+    for sampling_rate in $SAMPLING_RATE;do
+	for up_threshold in $UP_THRESHOLD;do
+	    command="${command} /var/log/cpufreq-bench/up_threshold_${up_threshold}_sampling_rate_${sampling_rate}/* \"up_threshold = $up_threshold - sampling_rate = $sampling_rate\""
+	done
+    done
+    echo "$command"
+    eval "$command"
+}
+
+measure
+create_plots
\ No newline at end of file

tools/power/cpupower/bench/example.cfg

+sleep = 50000
+load = 50000
+cpu = 0
+priority = LOW
+output = /var/log/cpufreq-bench
+sleep_step = 50000
+load_step = 50000
+cycles = 20
+rounds = 40
+verbose = 0
+governor = ondemand

tools/power/cpupower/bench/main.c

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <getopt.h>
+#include <errno.h>
+
+#include "config.h"
+#include "system.h"
+#include "benchmark.h"
+
+static struct option long_options[] = {
+	{"output",	1,	0,	'o'},
+	{"sleep",	1,	0,	's'},
+	{"load",	1,	0,	'l'},
+	{"verbose",	0,	0,	'v'},
+	{"cpu",		1,	0,	'c'},
+	{"governor",	1,	0,	'g'},
+	{"prio",	1,	0,	'p'},
+	{"file",	1,	0,	'f'},
+	{"cycles",	1,	0,	'n'},
+	{"rounds",	1,	0,	'r'},
+	{"load-step",	1,	0,	'x'},
+	{"sleep-step",	1,	0,	'y'},
+	{"help",	0,	0,	'h'},
+	{0, 0, 0, 0}
+};
+
+/*******************************************************************
+ usage
+*******************************************************************/
+
+void usage()
+{
+	printf("usage: ./bench\n");
+	printf("Options:\n");
+	printf(" -l, --load=<long int>\t\tinitial load time in us\n");
+	printf(" -s, --sleep=<long int>\t\tinitial sleep time in us\n");
+	printf(" -x, --load-step=<long int>\ttime to be added to load time, in us\n");
+	printf(" -y, --sleep-step=<long int>\ttime to be added to sleep time, in us\n");
+	printf(" -c, --cpu=<cpu #>\t\t\tCPU Nr. to use, starting at 0\n");
+	printf(" -p, --prio=<priority>\t\t\tscheduler priority, HIGH, LOW or DEFAULT\n");
+	printf(" -g, --governor=<governor>\t\tcpufreq governor to test\n");
+	printf(" -n, --cycles=<int>\t\t\tload/sleep cycles\n");
+	printf(" -r, --rounds<int>\t\t\tload/sleep rounds\n");
+	printf(" -f, --file=<configfile>\t\tconfig file to use\n");
+	printf(" -o, --output=<dir>\t\t\toutput path. Filename will be OUTPUTPATH/benchmark_TIMESTAMP.log\n");
+	printf(" -v, --verbose\t\t\t\tverbose output on/off\n");
+	printf(" -h, --help\t\t\t\tPrint this help screen\n");
+	exit(1);
+}
+
+/*******************************************************************
+ main
+*******************************************************************/
+
+int main(int argc, char **argv)
+{
+	int c;
+	int option_index = 0;
+	struct config *config = NULL;
+
+	config = prepare_default_config();
+
+	if (config == NULL)
+		return EXIT_FAILURE;
+
+	while (1) {
+		c = getopt_long (argc, argv, "hg:o:s:l:vc:p:f:n:r:x:y:",
+				long_options, &option_index);
+		if (c == -1)
+			break;
+
+		switch (c) {
+		case 'o':
+			if (config->output != NULL)
+				fclose(config->output);
+
+			config->output = prepare_output(optarg);
+
+			if (config->output == NULL)
+				return EXIT_FAILURE;
+
+			dprintf("user output path -> %s\n", optarg);
+			break;
+		case 's':
+			sscanf(optarg, "%li", &config->sleep);
+			dprintf("user sleep time -> %s\n", optarg);
+			break;
+		case 'l':
+			sscanf(optarg, "%li", &config->load);
+			dprintf("user load time -> %s\n", optarg);
+			break;
+		case 'c':
+			sscanf(optarg, "%u", &config->cpu);
+			dprintf("user cpu -> %s\n", optarg);
+			break;
+		case 'g':
+			strncpy(config->governor, optarg, 14);
+			dprintf("user governor -> %s\n", optarg);
+			break;
+		case 'p':
+			if (string_to_prio(optarg) != SCHED_ERR) {
+				config->prio = string_to_prio(optarg);
+				dprintf("user prio -> %s\n", optarg);
+			} else {
+				if (config != NULL) {
+					if (config->output != NULL)
+						fclose(config->output);
+					free(config);
+				}
+				usage();
+			}
+			break;
+		case 'n':
+			sscanf(optarg, "%u", &config->cycles);
+			dprintf("user cycles -> %s\n", optarg);
+			break;
+		case 'r':
+			sscanf(optarg, "%u", &config->rounds);
+			dprintf("user rounds -> %s\n", optarg);
+			break;
+		case 'x':
+			sscanf(optarg, "%li", &config->load_step);
+			dprintf("user load_step -> %s\n", optarg);
+			break;
+		case 'y':
+			sscanf(optarg, "%li", &config->sleep_step);
+			dprintf("user sleep_step -> %s\n", optarg);
+			break;
+		case 'f':
+			if (prepare_config(optarg, config))
+				return EXIT_FAILURE;
+			break;
+		case 'v':
+			config->verbose = 1;
+			dprintf("verbose output enabled\n");
+			break;
+		case 'h':
+		case '?':
+		default:
+			if (config != NULL) {
+				if (config->output != NULL)
+					fclose(config->output);
+				free(config);
+			}
+			usage();
+		}
+	}
+
+	if (config->verbose) {
+		printf("starting benchmark with parameters:\n");
+		printf("config:\n\t"
+		       "sleep=%li\n\t"
+		       "load=%li\n\t"
+		       "sleep_step=%li\n\t"
+		       "load_step=%li\n\t"
+		       "cpu=%u\n\t"
+		       "cycles=%u\n\t"
+		       "rounds=%u\n\t"
+		       "governor=%s\n\n",
+		       config->sleep,
+		       config->load,
+		       config->sleep_step,
+		       config->load_step,
+		       config->cpu,
+		       config->cycles,
+		       config->rounds,
+		       config->governor);
+	}
+
+	prepare_user(config);
+	prepare_system(config);
+	start_benchmark(config);
+
+	if (config->output != stdout)
+		fclose(config->output);
+
+	free(config);
+
+	return EXIT_SUCCESS;
+}
+

tools/power/cpupower/bench/parse.c

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <time.h>
+#include <dirent.h>
+
+#include <sys/utsname.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#include "parse.h"
+#include "config.h"
+
+/**
+ * converts priority string to priority
+ *
+ * @param str string that represents a scheduler priority
+ *
+ * @retval priority
+ * @retval SCHED_ERR when the priority doesn't exit
+ **/
+
+enum sched_prio string_to_prio(const char *str)
+{
+	if (strncasecmp("high", str, strlen(str)) == 0)
+		return  SCHED_HIGH;
+	else if (strncasecmp("default", str, strlen(str)) == 0)
+		return SCHED_DEFAULT;
+	else if (strncasecmp("low", str, strlen(str)) == 0)
+		return SCHED_LOW;
+	else
+		return SCHED_ERR;
+}
+
+/**
+ * create and open logfile
+ *
+ * @param dir directory in which the logfile should be created
+ *
+ * @retval logfile on success
+ * @retval NULL when the file can't be created
+ **/
+
+FILE *prepare_output(const char *dirname)
+{
+	FILE *output = NULL;
+	int len;
+	char *filename;
+	struct utsname sysdata;
+	DIR *dir;
+
+	dir = opendir(dirname);
+	if (dir == NULL) {
+		if (mkdir(dirname, 0755)) {
+			perror("mkdir");
+			fprintf(stderr, "error: Cannot create dir %s\n",
+				dirname);
+			return NULL;
+		}
+	}
+
+	len = strlen(dirname) + 30;
+	filename = malloc(sizeof(char) * len);
+
+	if (uname(&sysdata) == 0) {
+		len += strlen(sysdata.nodename) + strlen(sysdata.release);
+		filename = realloc(filename, sizeof(char) * len);
+
+		if (filename == NULL) {
+			perror("realloc");
+			return NULL;
+		}
+
+		snprintf(filename, len - 1, "%s/benchmark_%s_%s_%li.log",
+			dirname, sysdata.nodename, sysdata.release, time(NULL));
+	} else {
+		snprintf(filename, len - 1, "%s/benchmark_%li.log",
+			dirname, time(NULL));
+	}
+
+	dprintf("logilename: %s\n", filename);
+
+	output = fopen(filename, "w+");
+	if (output == NULL) {
+		perror("fopen");
+		fprintf(stderr, "error: unable to open logfile\n");
+	}
+
+	fprintf(stdout, "Logfile: %s\n", filename);
+
+	free(filename);
+	fprintf(output, "#round load sleep performance powersave percentage\n");
+	return output;
+}
+
+/**
+ * returns the default config
+ *
+ * @retval default config on success
+ * @retval NULL when the output file can't be created
+ **/
+
+struct config *prepare_default_config()
+{
+	struct config *config = malloc(sizeof(struct config));
+
+	dprintf("loading defaults\n");
+
+	config->sleep = 500000;
+	config->load = 500000;
+	config->sleep_step = 500000;
+	config->load_step = 500000;
+	config->cycles = 5;
+	config->rounds = 50;
+	config->cpu = 0;
+	config->prio = SCHED_HIGH;
+	config->verbose = 0;
+	strncpy(config->governor, "ondemand", 8);
+
+	config->output = stdout;
+
+#ifdef DEFAULT_CONFIG_FILE
+	if (prepare_config(DEFAULT_CONFIG_FILE, config))
+		return NULL;
+#endif
+	return config;
+}
+
+/**
+ * parses config file and returns the config to the caller
+ *
+ * @param path config file name
+ *
+ * @retval 1 on error
+ * @retval 0 on success
+ **/
+
+int prepare_config(const char *path, struct config *config)
+{
+	size_t len = 0;
+	char *opt, *val, *line = NULL;
+	FILE *configfile = fopen(path, "r");
+
+	if (config == NULL) {
+		fprintf(stderr, "error: config is NULL\n");
+		return 1;
+	}
+
+	if (configfile == NULL) {
+		perror("fopen");
+		fprintf(stderr, "error: unable to read configfile\n");
+		free(config);
+		return 1;
+	}
+
+	while (getline(&line, &len, configfile) != -1) {
+		if (line[0] == '#' || line[0] == ' ')
+			continue;
+
+		sscanf(line, "%as = %as", &opt, &val);
+
+		dprintf("parsing: %s -> %s\n", opt, val);
+
+		if (strncmp("sleep", opt, strlen(opt)) == 0)
+			sscanf(val, "%li", &config->sleep);
+
+		else if (strncmp("load", opt, strlen(opt)) == 0)
+			sscanf(val, "%li", &config->load);
+
+		else if (strncmp("load_step", opt, strlen(opt)) == 0)
+			sscanf(val, "%li", &config->load_step);
+
+		else if (strncmp("sleep_step", opt, strlen(opt)) == 0)
+			sscanf(val, "%li", &config->sleep_step);
+
+		else if (strncmp("cycles", opt, strlen(opt)) == 0)
+			sscanf(val, "%u", &config->cycles);
+
+		else if (strncmp("rounds", opt, strlen(opt)) == 0)
+			sscanf(val, "%u", &config->rounds);
+
+		else if (strncmp("verbose", opt, strlen(opt)) == 0)
+			sscanf(val, "%u", &config->verbose);
+
+		else if (strncmp("output", opt, strlen(opt)) == 0)
+			config->output = prepare_output(val); 
+
+		else if (strncmp("cpu", opt, strlen(opt)) == 0)
+			sscanf(val, "%u", &config->cpu);
+
+		else if (strncmp("governor", opt, 14) == 0)
+			strncpy(config->governor, val, 14);
+
+		else if (strncmp("priority", opt, strlen(opt)) == 0) {
+			if (string_to_prio(val) != SCHED_ERR)
+				config->prio = string_to_prio(val);
+		}
+	}
+
+	free(line);
+	free(opt);
+	free(val);
+
+	return 0;
+}

tools/power/cpupower/bench/parse.h

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+/* struct that holds the required config parameters */
+struct config
+{
+	long sleep;		/* sleep time in µs */
+	long load;		/* load time in µs */
+	long sleep_step;	/* time value which changes the
+				 * sleep time after every round in µs */
+	long load_step;		/* time value which changes the
+				 * load time after every round in µs */
+	unsigned int cycles;	/* calculation cycles with the same sleep/load time */
+	unsigned int rounds;	/* calculation rounds with iterated sleep/load time */
+	unsigned int cpu;	/* cpu for which the affinity is set */
+	char governor[15];	/* cpufreq governor */
+	enum sched_prio		/* possible scheduler priorities */
+	{
+		SCHED_ERR = -1,
+		SCHED_HIGH,
+		SCHED_DEFAULT,
+		SCHED_LOW
+	} prio;
+
+	unsigned int verbose;	/* verbose output */
+	FILE *output;		/* logfile */
+	char *output_filename;	/* logfile name, must be freed at the end
+				   if output != NULL and output != stdout*/
+};
+
+enum sched_prio string_to_prio(const char *str);
+
+FILE *prepare_output(const char *dir);
+
+int prepare_config(const char *path, struct config *config);
+struct config *prepare_default_config();
+

tools/power/cpupower/bench/system.c

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+#include <stdio.h>
+#include <time.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#include <sched.h>
+
+#include <cpufreq.h>
+
+#include "config.h"
+#include "system.h"
+
+/**
+ * returns time since epoch in µs
+ *
+ * @retval time
+ **/
+
+long long int get_time()
+{
+	struct timeval now;
+
+	gettimeofday(&now, NULL);
+
+	return (long long int)(now.tv_sec * 1000000LL + now.tv_usec);
+}
+
+/**
+ * sets the cpufreq governor
+ *
+ * @param governor cpufreq governor name
+ * @param cpu cpu for which the governor should be set
+ *
+ * @retval 0 on success
+ * @retval -1 when failed
+ **/
+
+int set_cpufreq_governor(char *governor, unsigned int cpu)
+{
+
+	dprintf("set %s as cpufreq governor\n", governor);
+
+	if (cpufreq_cpu_exists(cpu) != 0) {
+		perror("cpufreq_cpu_exists");
+		fprintf(stderr, "error: cpu %u does not exist\n", cpu);
+		return -1;
+	}
+
+	if (cpufreq_modify_policy_governor(cpu, governor) != 0) {
+		perror("cpufreq_modify_policy_governor");
+		fprintf(stderr, "error: unable to set %s governor\n", governor);
+		return -1;
+	}
+
+	return 0;
+}
+
+/**
+ * sets cpu affinity for the process
+ *
+ * @param cpu cpu# to which the affinity should be set
+ *
+ * @retval 0 on success
+ * @retval -1 when setting the affinity failed
+ **/
+
+int set_cpu_affinity(unsigned int cpu)
+{
+	cpu_set_t cpuset;
+
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+
+	dprintf("set affinity to cpu #%u\n", cpu);
+
+	if (sched_setaffinity(getpid(), sizeof(cpu_set_t), &cpuset) < 0) {
+		perror("sched_setaffinity");
+		fprintf(stderr, "warning: unable to set cpu affinity\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+/**
+ * sets the process priority parameter
+ *
+ * @param priority priority value
+ *
+ * @retval 0 on success
+ * @retval -1 when setting the priority failed
+ **/
+
+int set_process_priority(int priority)
+{
+	struct sched_param param;
+
+	dprintf("set scheduler priority to %i\n", priority);
+
+	param.sched_priority = priority;
+
+	if (sched_setscheduler(0, SCHEDULER, &param) < 0) {
+		perror("sched_setscheduler");
+		fprintf(stderr, "warning: unable to set scheduler priority\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+/**
+ * notifies the user that the benchmark may run some time
+ *
+ * @param config benchmark config values
+ *
+ **/
+
+void prepare_user(const struct config *config)
+{
+	unsigned long sleep_time = 0;
+	unsigned long load_time = 0;
+	unsigned int round;
+
+	for (round = 0; round < config->rounds; round++) {
+		sleep_time +=  2 * config->cycles *
+			(config->sleep + config->sleep_step * round);
+		load_time += 2 * config->cycles *
+			(config->load + config->load_step * round) +
+			(config->load + config->load_step * round * 4);
+	}
+
+	if (config->verbose || config->output != stdout)
+		printf("approx. test duration: %im\n",
+		       (int)((sleep_time + load_time) / 60000000));
+}
+
+/**
+ * sets up the cpu affinity and scheduler priority
+ *
+ * @param config benchmark config values
+ *
+ **/
+
+void prepare_system(const struct config *config)
+{
+	if (config->verbose)
+		printf("set cpu affinity to cpu #%u\n", config->cpu);
+
+	set_cpu_affinity(config->cpu);
+
+	switch (config->prio) {
+	case SCHED_HIGH:
+		if (config->verbose)
+			printf("high priority condition requested\n");
+
+		set_process_priority(PRIORITY_HIGH);
+		break;
+	case SCHED_LOW:
+		if (config->verbose)
+			printf("low priority condition requested\n");
+
+		set_process_priority(PRIORITY_LOW);
+		break;
+	default:
+		if (config->verbose)
+			printf("default priority condition requested\n");
+
+		set_process_priority(PRIORITY_DEFAULT);
+	}
+}
+

tools/power/cpupower/bench/system.h

+/*  cpufreq-bench CPUFreq microbenchmark
+ *
+ *  Copyright (C) 2008 Christian Kornacker <ckornacker@suse.de>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+
+#include "parse.h"
+
+long long get_time();
+
+int set_cpufreq_governor(char *governor, unsigned int cpu);
+int set_cpu_affinity(unsigned int cpu);
+int set_process_priority(int priority);
+
+void prepare_user(const struct config *config);
+void prepare_system(const struct config *config);

tools/power/cpupower/debug/i386/Makefile

+default: all
+
+centrino-decode: centrino-decode.c
+	$(CC) $(CFLAGS) -o centrino-decode centrino-decode.c
+
+dump_psb: dump_psb.c
+	$(CC) $(CFLAGS) -o dump_psb dump_psb.c
+
+intel_gsic: intel_gsic.c
+	$(CC) $(CFLAGS) -o intel_gsic -llrmi intel_gsic.c
+
+powernow-k8-decode: powernow-k8-decode.c
+	$(CC) $(CFLAGS) -o powernow-k8-decode powernow-k8-decode.c
+
+all: centrino-decode dump_psb intel_gsic powernow-k8-decode
+
+clean:
+	rm -rf centrino-decode dump_psb intel_gsic powernow-k8-decode
+
+.PHONY: all default clean

tools/power/cpupower/debug/i386/centrino-decode.c

+/*
+ *  (C) 2003 - 2004  Dominik Brodowski <linux@dominikbrodowski.de>
+ *
+ *  Licensed under the terms of the GNU GPL License version 2.
+ *
+ * Based on code found in
+ * linux/arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c
+ * and originally developed by Jeremy Fitzhardinge.
+ *
+ * USAGE: simply run it to decode the current settings on CPU 0,
+ *	  or pass the CPU number as argument, or pass the MSR content
+ *	  as argument.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <fcntl.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#define MCPU	32
+
+#define MSR_IA32_PERF_STATUS	0x198
+
+static int rdmsr(unsigned int cpu, unsigned int msr,
+		 unsigned int *lo, unsigned int *hi)
+{
+	int fd;
+	char file[20];
+	unsigned long long val;
+	int retval = -1;
+
+	*lo = *hi = 0;
+
+	if (cpu > MCPU)
+		goto err1;
+
+	sprintf(file, "/dev/cpu/%d/msr", cpu);
+	fd = open(file, O_RDONLY);
+
+	if (fd < 0)
+		goto err1;
+
+	if (lseek(fd, msr, SEEK_CUR) == -1)
+		goto err2;
+
+	if (read(fd, &val, 8) != 8)
+		goto err2;
+
+	*lo = (uint32_t )(val & 0xffffffffull);
+	*hi = (uint32_t )(val>>32 & 0xffffffffull);
+
+	retval = 0;
+err2:
+	close(fd);
+err1:
+	return retval;
+}
+
+static void decode (unsigned int msr)
+{
+	unsigned int multiplier;
+	unsigned int mv;
+
+	multiplier = ((msr >> 8) & 0xFF);
+
+	mv = (((msr & 0xFF) * 16) + 700);
+
+	printf("0x%x means multiplier %d @ %d mV\n", msr, multiplier, mv);
+}
+
+static int decode_live(unsigned int cpu)
+{
+	unsigned int lo, hi;
+	int err;
+
+	err = rdmsr(cpu, MSR_IA32_PERF_STATUS, &lo, &hi);
+
+	if (err) {
+		printf("can't get MSR_IA32_PERF_STATUS for cpu %d\n", cpu);
+		printf("Possible trouble: you don't run an Enhanced SpeedStep capable cpu\n");
+		printf("or you are not root, or the msr driver is not present\n");
+		return 1;
+	}
+
+	decode(lo);
+
+	return 0;
+}
+
+int main (int argc, char **argv)
+{
+	unsigned int cpu, mode = 0;
+
+	if (argc < 2)
+		cpu = 0;
+	else {
+		cpu = strtoul(argv[1], NULL, 0);
+		if (cpu >= MCPU)
+			mode = 1;
+	}
+
+	if (mode)
+		decode(cpu);
+	else
+		decode_live(cpu);
+
+	return 0;
+}

tools/power/cpupower/debug/i386/dump_psb.c

+/*
+ * dump_psb. (c) 2004, Dave Jones, Red Hat Inc.
+ * Licensed under the GPL v2.
+ */
+
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define _GNU_SOURCE
+#include <getopt.h>
+
+#include <sys/mman.h>
+
+#define LEN (0x100000 - 0xc0000)
+#define OFFSET (0xc0000)
+
+#ifndef __packed
+#define __packed __attribute((packed))
+#endif
+
+static long relevant;
+
+static const int fid_to_mult[32] = {
+	110, 115, 120, 125, 50, 55, 60, 65,
+	70, 75, 80, 85, 90, 95, 100, 105,
+	30, 190, 40, 200, 130, 135, 140, 210,
+	150, 225, 160, 165, 170, 180, -1, -1,
+};
+
+static const int vid_to_voltage[32] = {
+	2000, 1950, 1900, 1850, 1800, 1750, 1700, 1650,
+	1600, 1550, 1500, 1450, 1400, 1350, 1300, 0,
+	1275, 1250, 1225, 1200, 1175, 1150, 1125, 1100,
+	1075, 1050, 1024, 1000, 975, 950, 925, 0,
+};
+
+struct psb_header {
+	char signature[10];
+	u_char version;
+	u_char flags;
+	u_short settlingtime;
+	u_char res1;
+	u_char numpst;
+} __packed;
+
+struct pst_header {
+	u_int32_t cpuid;
+	u_char fsb;
+	u_char maxfid;
+	u_char startvid;
+	u_char numpstates;
+} __packed;
+
+static u_int fsb;
+static u_int sgtc;
+
+static int
+decode_pst(char *p, int npstates)
+{
+	int i;
+	int freq, fid, vid;
+
+	for (i = 0; i < npstates; ++i) {
+		fid = *p++;
+		vid = *p++;
+		freq = 100 * fid_to_mult[fid] * fsb;
+
+		printf("   %2d %8dkHz  FID %02x (%2d.%01d)  VID %02x (%4dmV)\n",
+		       i,
+		       freq,
+		       fid, fid_to_mult[fid]/10, fid_to_mult[fid]%10,
+		       vid, vid_to_voltage[vid]);
+	}
+
+	return 0;
+}
+
+static
+void decode_psb(char *p, int numpst)
+{
+	int i;
+	struct psb_header *psb;
+	struct pst_header *pst;
+
+	psb = (struct psb_header*) p;
+
+	if (psb->version != 0x12)
+		return;
+
+	printf("PSB version: %hhx flags: %hhx settling time %hhuus res1 %hhx num pst %hhu\n",
+			psb->version,
+			psb->flags,
+			psb->settlingtime,
+			psb->res1,
+			psb->numpst);
+	sgtc = psb->settlingtime * 100;
+
+	if (sgtc < 10000)
+		sgtc = 10000;
+
+	p = ((char *) psb) + sizeof(struct psb_header);
+
+	if (numpst < 0)
+		numpst = psb->numpst;
+	else
+		printf("Overriding number of pst :%d\n", numpst);
+
+	for (i = 0; i < numpst; i++) {
+		pst = (struct pst_header*) p;
+
+		if (relevant != 0) {
+			if (relevant!= pst->cpuid)
+				goto next_one;
+		}
+
+		printf("  PST %d  cpuid %.3x fsb %hhu mfid %hhx svid %hhx numberstates %hhu\n",
+				i+1,
+				pst->cpuid,
+				pst->fsb,
+				pst->maxfid,
+				pst->startvid,
+				pst->numpstates);
+
+		fsb = pst->fsb;
+		decode_pst(p + sizeof(struct pst_header), pst->numpstates);
+
+next_one:
+		p += sizeof(struct pst_header) + 2*pst->numpstates;
+	}
+
+}
+
+static struct option info_opts[] = {
+	{.name = "numpst",	.has_arg=no_argument,	.flag=NULL, .val='n'},
+};
+
+void print_help(void)
+{
+	printf ("Usage: dump_psb [options]\n");
+	printf ("Options:\n");
+	printf ("  -n, --numpst     Set number of PST tables to scan\n");
+	printf ("  -r, --relevant   Only display PSTs relevant to cpuid N\n");
+}
+
+int
+main(int argc, char *argv[])
+{
+	int fd;
+	int numpst=-1;
+	int ret=0, cont=1;
+	char *mem = NULL;
+	char *p;
+
+	do {
+		ret = getopt_long(argc, argv, "hr:n:", info_opts, NULL);
+		switch (ret){
+		case '?':
+		case 'h':
+			print_help();
+			cont = 0;
+			break;
+		case 'r':
+			relevant = strtol(optarg, NULL, 16);
+			break;
+		case 'n':
+			numpst = strtol(optarg, NULL, 10);
+			break;
+		case -1:
+			cont = 0;
+			break;
+		}
+
+	} while(cont);
+
+	fd = open("/dev/mem", O_RDONLY);
+	if (fd < 0) {
+		printf ("Couldn't open /dev/mem. Are you root?\n");
+		exit(1);
+	}
+
+	mem = mmap(mem, 0x100000 - 0xc0000, PROT_READ, MAP_SHARED, fd, 0xc0000);
+	close(fd);
+
+	for (p = mem; p - mem < LEN; p+=16) {
+		if (memcmp(p, "AMDK7PNOW!", 10) == 0) {
+			decode_psb(p, numpst);
+			break;
+		}
+	}
+
+	munmap(mem, LEN);
+	return 0;
+}

tools/power/cpupower/debug/i386/intel_gsic.c

+/*
+ *  (C) 2003  Bruno Ducrot
+ *  (C) 2004  Dominik Brodowski <linux@dominikbrodowski.de>
+ *
+ *  Licensed under the terms of the GNU GPL License version 2.
+ *
+ * Based on code found in
+ * linux/include/asm-i386/ist.h and linux/arch/i386/kernel/setup.c
+ * and originally developed by Andy Grover <andrew.grover@intel.com>
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <lrmi.h>
+
+int main (void)
+{
+	struct LRMI_regs	r;
+	int			retval;
+
+	if (!LRMI_init())
+		return 0;
+
+	memset(&r, 0, sizeof(r));
+
+	r.eax = 0x0000E980;
+	r.edx = 0x47534943;
+
+	retval = LRMI_int(0x15, &r);
+
+	if (!retval) {
+		printf("Failed!\n");
+		return 0;
+	}
+	if (r.eax == 0x47534943) {
+		printf("BIOS supports GSIC call:\n");
+		printf("\tsignature: %c%c%c%c\n",
+		       (r.eax >> 24) & 0xff,
+		       (r.eax >> 16) & 0xff,
+		       (r.eax >> 8) & 0xff,
+		       (r.eax) & 0xff);
+		printf("\tcommand port = 0x%.4x\n",
+		       r.ebx & 0xffff);
+		printf("\tcommand =      0x%.4x\n",
+		       (r.ebx >> 16) & 0xffff);
+		printf("\tevent port =   0x%.8x\n", r.ecx);
+		printf("\tflags =        0x%.8x\n", r.edx);
+		if (((r.ebx >> 16) & 0xffff) != 0x82) {
+			printf("non-default command value. If speedstep-smi "
+			       "doesn't work out of the box,\nyou may want to "
+			       "try out the default value by passing "
+			       "smi_cmd=0x82 to the module\n ON YOUR OWN "
+			       "RISK.\n");
+		}
+		if ((r.ebx & 0xffff) != 0xb2) {
+			printf("non-default command port. If speedstep-smi "
+			       "doesn't work out of the box,\nyou may want to "
+			       "try out the default value by passing "
+			       "smi_port=0x82 to the module\n ON YOUR OWN "
+			       "RISK.\n");
+		}
+	} else {
+		printf("BIOS DOES NOT support GSIC call.  Dumping registers anyway:\n");
+		printf("eax = 0x%.8x\n", r.eax);
+		printf("ebx = 0x%.8x\n", r.ebx);
+		printf("ecx = 0x%.8x\n", r.ecx);
+		printf("edx = 0x%.8x\n", r.edx);
+		printf("Note also that some BIOS do not support the initial "
+		       "GSIC call, but the newer\nspeeedstep-smi driver may "
+		       "work.\nFor this, you need to pass some arguments to "
+		       "the speedstep-smi driver:\n");
+		printf("\tsmi_cmd=0x?? smi_port=0x?? smi_sig=1\n");
+		printf("\nUnfortunately, you have to know what exactly are "
+		       "smi_cmd and smi_port, and this\nis system "
+		       "dependant.\n");
+	}
+	return 1;
+}

tools/power/cpupower/debug/i386/powernow-k8-decode.c

+/*
+ *  (C) 2004 Bruno Ducrot <ducrot@poupinou.org>
+ *
+ *  Licensed under the terms of the GNU GPL License version 2.
+ *
+ * Based on code found in
+ * linux/arch/i386/kernel/cpu/cpufreq/powernow-k8.c
+ * and originally developed by Paul Devriendt
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <fcntl.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#define MCPU 32
+
+#define MSR_FIDVID_STATUS	0xc0010042
+
+#define MSR_S_HI_CURRENT_VID	0x0000001f
+#define MSR_S_LO_CURRENT_FID	0x0000003f
+
+static int get_fidvid(uint32_t cpu, uint32_t *fid, uint32_t *vid)
+{
+	int err = 1;
+	uint64_t msr = 0;
+	int fd;
+	char file[20];
+
+	if (cpu > MCPU)
+		goto out;
+
+	sprintf(file, "/dev/cpu/%d/msr", cpu);
+
+	fd = open(file, O_RDONLY);
+	if (fd < 0)
+		goto out;
+	lseek(fd, MSR_FIDVID_STATUS, SEEK_CUR);
+	if (read(fd, &msr, 8) != 8)
+		goto err1;
+
+	*fid = ((uint32_t )(msr & 0xffffffffull)) & MSR_S_LO_CURRENT_FID;
+	*vid = ((uint32_t )(msr>>32 & 0xffffffffull)) & MSR_S_HI_CURRENT_VID;
+	err = 0;
+err1:
+	close(fd);
+out:
+	return err;
+}
+
+
+/* Return a frequency in MHz, given an input fid */
+static uint32_t find_freq_from_fid(uint32_t fid)
+{
+	return 800 + (fid * 100);
+}
+
+/* Return a voltage in miliVolts, given an input vid */
+static uint32_t find_millivolts_from_vid(uint32_t vid)
+{
+	return 1550-vid*25;
+}
+
+int main (int argc, char *argv[])
+{
+	int err;
+	int cpu;
+	uint32_t fid, vid;
+
+	if (argc < 2)
+		cpu = 0;
+	else
+		cpu = strtoul(argv[1], NULL, 0);
+
+	err = get_fidvid(cpu, &fid, &vid);
+
+	if (err) {
+		printf("can't get fid, vid from MSR\n");
+		printf("Possible trouble: you don't run a powernow-k8 capable cpu\n");
+		printf("or you are not root, or the msr driver is not present\n");
+		exit(1);
+	}
+
+	
+	printf("cpu %d currently at %d MHz and %d mV\n",
+			cpu,
+			find_freq_from_fid(fid),
+			find_millivolts_from_vid(vid));
+	
+	return 0;
+}

tools/power/cpupower/debug/kernel/Makefile

+obj-m	:=
+
+KDIR	:= /lib/modules/$(shell uname -r)/build
+PWD		:= $(shell pwd)
+KMISC   := /lib/modules/$(shell uname -r)/cpufrequtils/
+
+ifeq ("$(CONFIG_X86_TSC)", "y")
+  obj-m	 += cpufreq-test_tsc.o
+endif
+
+default:
+	$(MAKE) -C $(KDIR) M=$(PWD)
+
+clean:
+	- rm -rf *.o *.ko .tmp-versions .*.cmd .*.mod.* *.mod.c
+	- rm -rf .tmp_versions* Module.symvers modules.order
+
+install: default
+	install -d $(KMISC)
+	install -m 644 -c *.ko $(KMISC)
+	/sbin/depmod -a
+
+all:	default

tools/power/cpupower/debug/kernel/cpufreq-test_tsc.c

+/*
+ * test module to check whether the TSC-based delay routine continues
+ * to work properly after cpufreq transitions. Needs ACPI to work
+ * properly.
+ *
+ * Based partly on the Power Management Timer (PMTMR) code to be found
+ * in arch/i386/kernel/timers/timer_pm.c on recent 2.6. kernels, especially
+ * code written by John Stultz. The read_pmtmr function was copied verbatim
+ * from that file.
+ *
+ * (C) 2004 Dominik Brodowski
+ *
+ * To use:
+ * 1.) pass clock=tsc to the kernel on your bootloader
+ * 2.) modprobe this module (it'll fail)
+ * 3.) change CPU frequency
+ * 4.) modprobe this module again
+ * 5.) if the third value, "diff_pmtmr", changes between 2. and 4., the
+ *     TSC-based delay routine on the Linux kernel does not correctly
+ *     handle the cpufreq transition. Please report this to
+ *     cpufreq@vger.kernel.org
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+
+#include <asm/io.h>
+
+#include <acpi/acpi_bus.h>
+#include <acpi/acpi_drivers.h>
+
+static int pm_tmr_ioport = 0;
+
+/*helper function to safely read acpi pm timesource*/
+static u32 read_pmtmr(void)
+{
+	u32 v1=0,v2=0,v3=0;
+	/* It has been reported that because of various broken
+	 * chipsets (ICH4, PIIX4 and PIIX4E) where the ACPI PM time
+	 * source is not latched, so you must read it multiple
+	 * times to insure a safe value is read.
+	 */
+	do {
+		v1 = inl(pm_tmr_ioport);
+		v2 = inl(pm_tmr_ioport);
+		v3 = inl(pm_tmr_ioport);
+	} while ((v1 > v2 && v1 < v3) || (v2 > v3 && v2 < v1)
+		 || (v3 > v1 && v3 < v2));
+
+	/* mask the output to 24 bits */
+	return (v2 & 0xFFFFFF);
+}
+
+static int __init cpufreq_test_tsc(void)
+{
+	u32 now, then, diff;
+	u64 now_tsc, then_tsc, diff_tsc;
+	int i;
+
+	/* the following code snipped is copied from arch/x86/kernel/acpi/boot.c
+	   of Linux v2.6.25. */
+
+	/* detect the location of the ACPI PM Timer */
+	if (acpi_gbl_FADT.header.revision >= FADT2_REVISION_ID) {
+		/* FADT rev. 2 */
+		if (acpi_gbl_FADT.xpm_timer_block.space_id !=
+		    ACPI_ADR_SPACE_SYSTEM_IO)
+			return 0;
+
+		pm_tmr_ioport = acpi_gbl_FADT.xpm_timer_block.address;
+		/*
+		 * "X" fields are optional extensions to the original V1.0
+		 * fields, so we must selectively expand V1.0 fields if the
+		 * corresponding X field is zero.
+	 	 */
+		if (!pm_tmr_ioport)
+			pm_tmr_ioport = acpi_gbl_FADT.pm_timer_block;
+	} else {
+		/* FADT rev. 1 */
+		pm_tmr_ioport = acpi_gbl_FADT.pm_timer_block;
+	}
+
+	printk(KERN_DEBUG "start--> \n");
+	then = read_pmtmr();
+        rdtscll(then_tsc);
+	for (i=0;i<20;i++) {
+		mdelay(100);
+		now = read_pmtmr();
+		rdtscll(now_tsc);
+		diff = (now - then) & 0xFFFFFF;
+		diff_tsc = now_tsc - then_tsc;
+		printk(KERN_DEBUG "t1: %08u t2: %08u diff_pmtmr: %08u diff_tsc: %016llu\n", then, now, diff, diff_tsc);
+		then = now;
+		then_tsc = now_tsc;
+	}
+	printk(KERN_DEBUG "<-- end \n");
+	return -ENODEV;
+}
+
+static void __exit cpufreq_none(void)
+{
+	return;
+}
+
+module_init(cpufreq_test_tsc)
+module_exit(cpufreq_none)
+
+
+MODULE_AUTHOR("Dominik Brodowski");
+MODULE_DESCRIPTION("Verify the TSC cpufreq notifier working correctly -- needs ACPI-enabled system");
+MODULE_LICENSE ("GPL");

tools/power/cpupower/debug/x86_64/Makefile

+default: all
+
+centrino-decode: centrino-decode.c
+	$(CC) $(CFLAGS) -o centrino-decode centrino-decode.c
+
+powernow-k8-decode: powernow-k8-decode.c
+	$(CC) $(CFLAGS) -o powernow-k8-decode powernow-k8-decode.c
+
+all: centrino-decode powernow-k8-decode
+
+clean:
+	rm -rf centrino-decode powernow-k8-decode
+
+.PHONY: all default clean

tools/power/cpupower/debug/x86_64/centrino-decode.c

+../i386/centrino-decode.c
\ No newline at end of file

tools/power/cpupower/debug/x86_64/powernow-k8-decode.c

+../i386/powernow-k8-decode.c
\ No newline at end of file

tools/power/cpupower/lib/cpufreq.c

+/*
+ *  (C) 2004-2009  Dominik Brodowski <linux@dominikbrodowski.de>
+ *
+ *  Licensed under the terms of the GNU GPL License version 2.
+ */
+
+
+#include <stdio.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "cpufreq.h"
+#include "sysfs.h"
+
+int cpufreq_cpu_exists(unsigned int cpu)
+{
+	return sysfs_cpu_exists(cpu);
+}
+
+unsigned long cpufreq_get_freq_kernel(unsigned int cpu)
+{
+	return sysfs_get_freq_kernel(cpu);
+}
+
+unsigned long cpufreq_get_freq_hardware(unsigned int cpu)
+{
+	return sysfs_get_freq_hardware(cpu);
+}
+
+unsigned long cpufreq_get_transition_latency(unsigned int cpu)
+{
+	return sysfs_get_freq_transition_latency(cpu);
+}
+
+int cpufreq_get_hardware_limits(unsigned int cpu,
+				unsigned long *min,
+				unsigned long *max)
+{
+	if ((!min) || (!max))
+		return -EINVAL;
+	return sysfs_get_freq_hardware_limits(cpu, min, max);
+}
+
+char *cpufreq_get_driver(unsigned int cpu)
+{
+	return sysfs_get_freq_driver(cpu);
+}
+
+void cpufreq_put_driver(char *ptr)
+{
+	if (!ptr)
+		return;
+	free(ptr);
+}
+
+struct cpufreq_policy *cpufreq_get_policy(unsigned int cpu)
+{
+	return sysfs_get_freq_policy(cpu);
+}
+
+void cpufreq_put_policy(struct cpufreq_policy *policy)
+{
+	if ((!policy) || (!policy->governor))
+		return;
+
+	free(policy->governor);
+	policy->governor = NULL;
+	free(policy);
+}
+
+struct cpufreq_available_governors *cpufreq_get_available_governors(unsigned
+								int cpu)
+{
+	return sysfs_get_freq_available_governors(cpu);
+}
+
+void cpufreq_put_available_governors(struct cpufreq_available_governors *any)
+{
+	struct cpufreq_available_governors *tmp, *next;
+
+	if (!any)
+		return;
+
+	tmp = any->first;
+	while (tmp) {
+		next = tmp->next;
+		if (tmp->governor)
+			free(tmp->governor);
+		free(tmp);
+		tmp = next;
+	}
+}
+
+
+struct cpufreq_available_frequencies
+*cpufreq_get_available_frequencies(unsigned int cpu)
+{
+	return sysfs_get_available_frequencies(cpu);
+}
+
+void cpufreq_put_available_frequencies(struct cpufreq_available_frequencies
+				*any) {
+	struct cpufreq_available_frequencies *tmp, *next;
+
+	if (!any)
+		return;
+
+	tmp = any->first;
+	while (tmp) {
+		next = tmp->next;
+		free(tmp);
+		tmp = next;
+	}
+}
+
+
+struct cpufreq_affected_cpus *cpufreq_get_affected_cpus(unsigned int cpu)
+{
+	return sysfs_get_freq_affected_cpus(cpu);
+}
+
+void cpufreq_put_affected_cpus(struct cpufreq_affected_cpus *any)
+{
+	struct cpufreq_affected_cpus *tmp, *next;
+
+	if (!any)
+		return;
+
+	tmp = any->first;
+	while (tmp) {
+		next = tmp->next;
+		free(tmp);
+		tmp = next;
+	}
+}
+
+
+struct cpufreq_affected_cpus *cpufreq_get_related_cpus(unsigned int cpu)
+{
+	return sysfs_get_freq_related_cpus(cpu);
+}
+
+void cpufreq_put_related_cpus(struct cpufreq_affected_cpus *any)
+{
+	cpufreq_put_affected_cpus(any);
+}
+
+
+int cpufreq_set_policy(unsigned int cpu, struct cpufreq_policy *policy)
+{
+	if (!policy || !(policy->governor))
+		return -EINVAL;
+
+	return sysfs_set_freq_policy(cpu, policy);
+}
+
+
+int cpufreq_modify_policy_min(unsigned int cpu, unsigned long min_freq)
+{
+	return sysfs_modify_freq_policy_min(cpu, min_freq);
+}
+
+
+int cpufreq_modify_policy_max(unsigned int cpu, unsigned long max_freq)
+{
+	return sysfs_modify_freq_policy_max(cpu, max_freq);
+}
+
+
+int cpufreq_modify_policy_governor(unsigned int cpu, char *governor)
+{
+	if ((!governor) || (strlen(governor) > 19))
+		return -EINVAL;
+
+	return sysfs_modify_freq_policy_governor(cpu, governor);
+}
+
+int cpufreq_set_frequency(unsigned int cpu, unsigned long target_frequency)
+{
+	return sysfs_set_frequency(cpu, target_frequency);
+}
+
+struct cpufreq_stats *cpufreq_get_stats(unsigned int cpu,
+					unsigned long long *total_time)
+{
+	return sysfs_get_freq_stats(cpu, total_time);
+}
+
+void cpufreq_put_stats(struct cpufreq_stats *any)
+{
+	struct cpufreq_stats *tmp, *next;
+
+	if (!any)
+		return;
+
+	tmp = any->first;
+	while (tmp) {
+		next = tmp->next;
+		free(tmp);
+		tmp = next;
+	}
+}
+
+unsigned long cpufreq_get_transitions(unsigned int cpu)
+{
+	return sysfs_get_freq_transitions(cpu);
+}

tools/power/cpupower/lib/cpufreq.h

+/*
+ *  cpufreq.h - definitions for libcpufreq
+ *
+ *  Copyright (C) 2004-2009  Dominik Brodowski <linux@dominikbrodowski.de>
+ *
+ *  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; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _CPUFREQ_H
+#define _CPUFREQ_H 1
+
+struct cpufreq_policy {
+	unsigned long min;
+	unsigned long max;
+	char *governor;
+};
+
+struct cpufreq_available_governors {
+	char *governor;
+	struct cpufreq_available_governors *next;
+	struct cpufreq_available_governors *first;
+};
+
+struct cpufreq_available_frequencies {
+	unsigned long frequency;
+	struct cpufreq_available_frequencies *next;
+	struct cpufreq_available_frequencies *first;
+};
+
+
+struct cpufreq_affected_cpus {
+	unsigned int cpu;
+	struct cpufreq_affected_cpus *next;
+	struct cpufreq_affected_cpus *first;
+};
+
+struct cpufreq_stats {
+	unsigned long frequency;
+	unsigned long long time_in_state;
+	struct cpufreq_stats *next;
+	struct cpufreq_stats *first;
+};
+
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * returns 0 if the specified CPU is present (it doesn't say
+ * whether it is online!), and an error value if not.
+ */
+
+extern int cpufreq_cpu_exists(unsigned int cpu);
+
+/* determine current CPU frequency
+ * - _kernel variant means kernel's opinion of CPU frequency
+ * - _hardware variant means actual hardware CPU frequency,
+ *    which is only available to root.
+ *
+ * returns 0 on failure, else frequency in kHz.
+ */
+
+extern unsigned long cpufreq_get_freq_kernel(unsigned int cpu);
+
+extern unsigned long cpufreq_get_freq_hardware(unsigned int cpu);
+
+#define cpufreq_get(cpu) cpufreq_get_freq_kernel(cpu);
+
+
+/* determine CPU transition latency
+ *
+ * returns 0 on failure, else transition latency in 10^(-9) s = nanoseconds
+ */
+extern unsigned long cpufreq_get_transition_latency(unsigned int cpu);
+
+
+/* determine hardware CPU frequency limits
+ *
+ * These may be limited further by thermal, energy or other
+ * considerations by cpufreq policy notifiers in the kernel.
+ */
+
+extern int cpufreq_get_hardware_limits(unsigned int cpu,
+				unsigned long *min,
+				unsigned long *max);
+
+
+/* determine CPUfreq driver used
+ *
+ * Remember to call cpufreq_put_driver when no longer needed
+ * to avoid memory leakage, please.
+ */
+
+extern char *cpufreq_get_driver(unsigned int cpu);
+
+extern void cpufreq_put_driver(char *ptr);
+
+
+/* determine CPUfreq policy currently used
+ *
+ * Remember to call cpufreq_put_policy when no longer needed
+ * to avoid memory leakage, please.
+ */
+
+
+extern struct cpufreq_policy *cpufreq_get_policy(unsigned int cpu);
+
+extern void cpufreq_put_policy(struct cpufreq_policy *policy);
+
+
+/* determine CPUfreq governors currently available
+ *
+ * may be modified by modprobe'ing or rmmod'ing other governors. Please
+ * free allocated memory by calling cpufreq_put_available_governors
+ * after use.
+ */
+
+
+extern struct cpufreq_available_governors
+*cpufreq_get_available_governors(unsigned int cpu);
+
+extern void cpufreq_put_available_governors(
+	struct cpufreq_available_governors *first);
+
+
+/* determine CPU frequency states available
+ *
+ * Only present on _some_ ->target() cpufreq drivers. For information purposes
+ * only. Please free allocated memory by calling
+ * cpufreq_put_available_frequencies after use.
+ */
+
+extern struct cpufreq_available_frequencies
+*cpufreq_get_available_frequencies(unsigned int cpu);
+
+extern void cpufreq_put_available_frequencies(
+		struct cpufreq_available_frequencies *first);
+
+
+/* determine affected CPUs
+ *
+ * Remember to call cpufreq_put_affected_cpus when no longer needed
+ * to avoid memory leakage, please.
+ */
+
+extern struct cpufreq_affected_cpus *cpufreq_get_affected_cpus(unsigned
+							int cpu);
+
+extern void cpufreq_put_affected_cpus(struct cpufreq_affected_cpus *first);
+
+
+/* determine related CPUs
+ *
+ * Remember to call cpufreq_put_related_cpus when no longer needed
+ * to avoid memory leakage, please.
+ */
+
+extern struct cpufreq_affected_cpus *cpufreq_get_related_cpus(unsigned
+							int cpu);
+
+extern void cpufreq_put_related_cpus(struct cpufreq_affected_cpus *first);
+
+
+/* determine stats for cpufreq subsystem
+ *
+ * This is not available in all kernel versions or configurations.
+ */
+
+extern struct cpufreq_stats *cpufreq_get_stats(unsigned int cpu,
+					unsigned long long *total_time);
+
+extern void cpufreq_put_stats(struct cpufreq_stats *stats);
+
+extern unsigned long cpufreq_get_transitions(unsigned int cpu);
+
+
+/* set new cpufreq policy
+ *
+ * Tries to set the passed policy as new policy as close as possible,
+ * but results may differ depending e.g. on governors being available.
+ */
+
+extern int cpufreq_set_policy(unsigned int cpu, struct cpufreq_policy *policy);
+
+
+/* modify a policy by only changing min/max freq or governor
+ *
+ * Does not check whether result is what was intended.
+ */
+
+extern int cpufreq_modify_policy_min(unsigned int cpu, unsigned long min_freq);
+extern int cpufreq_modify_policy_max(unsigned int cpu, unsigned long max_freq);
+extern int cpufreq_modify_policy_governor(unsigned int cpu, char *governor);
+
+
+/* set a specific frequency
+ *
+ * Does only work if userspace governor can be used and no external
+ * interference (other calls to this function or to set/modify_policy)
+ * occurs. Also does not work on ->range() cpufreq drivers.
+ */
+
+extern int cpufreq_set_frequency(unsigned int cpu,
+				unsigned long target_frequency);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _CPUFREQ_H */

tools/power/cpupower/lib/sysfs.h

+/* General */
+extern unsigned int sysfs_cpu_exists(unsigned int cpu);
+
+/* CPUfreq */
+extern unsigned long sysfs_get_freq_kernel(unsigned int cpu);
+extern unsigned long sysfs_get_freq_hardware(unsigned int cpu);
+extern unsigned long sysfs_get_freq_transition_latency(unsigned int cpu);
+extern int sysfs_get_freq_hardware_limits(unsigned int cpu,
+					unsigned long *min, unsigned long *max);
+extern char *sysfs_get_freq_driver(unsigned int cpu);
+extern struct cpufreq_policy *sysfs_get_freq_policy(unsigned int cpu);
+extern struct cpufreq_available_governors *sysfs_get_freq_available_governors(
+	unsigned int cpu);
+extern struct cpufreq_available_frequencies *sysfs_get_available_frequencies(
+	unsigned int cpu);
+extern struct cpufreq_affected_cpus *sysfs_get_freq_affected_cpus(
+	unsigned int cpu);
+extern struct cpufreq_affected_cpus *sysfs_get_freq_related_cpus(
+	unsigned int cpu);
+extern struct cpufreq_stats *sysfs_get_freq_stats(unsigned int cpu,
+						unsigned long long *total_time);
+extern unsigned long sysfs_get_freq_transitions(unsigned int cpu);
+extern int sysfs_set_freq_policy(unsigned int cpu,
+				struct cpufreq_policy *policy);
+extern int sysfs_modify_freq_policy_min(unsigned int cpu,
+					unsigned long min_freq);
+extern int sysfs_modify_freq_policy_max(unsigned int cpu,
+					unsigned long max_freq);
+extern int sysfs_modify_freq_policy_governor(unsigned int cpu, char *governor);
+extern int sysfs_set_frequency(unsigned int cpu,
+			unsigned long target_frequency);

tools/power/cpupower/man/cpupower-frequency-info.1

+.TH "cpufreq-info" "1" "0.1" "Mattia Dongili" ""
+.SH "NAME"
+.LP 
+cpufreq\-info \- Utility to retrieve cpufreq kernel information
+.SH "SYNTAX"
+.LP 
+cpufreq\-info [\fIoptions\fP]
+.SH "DESCRIPTION"
+.LP 
+A small tool which prints out cpufreq information helpful to developers and interested users.
+.SH "OPTIONS"
+.LP 
+.TP  
+\fB\-e\fR \fB\-\-debug\fR
+Prints out debug information.
+.TP  
+\fB\-f\fR \fB\-\-freq\fR
+Get frequency the CPU currently runs at, according to the cpufreq core.
+.TP  
+\fB\-w\fR \fB\-\-hwfreq\fR
+Get frequency the CPU currently runs at, by reading it from hardware (only available to root).
+.TP  
+\fB\-l\fR \fB\-\-hwlimits\fR
+Determine the minimum and maximum CPU frequency allowed.
+.TP  
+\fB\-d\fR \fB\-\-driver\fR
+Determines the used cpufreq kernel driver.
+.TP  
+\fB\-p\fR \fB\-\-policy\fR
+Gets the currently used cpufreq policy.
+.TP  
+\fB\-g\fR \fB\-\-governors\fR
+Determines available cpufreq governors.
+.TP  
+\fB\-a\fR \fB\-\-related\-cpus\fR
+Determines which CPUs run at the same hardware frequency.
+.TP  
+\fB\-a\fR \fB\-\-affected\-cpus\fR
+Determines which CPUs need to have their frequency coordinated by software.
+.TP  
+\fB\-s\fR \fB\-\-stats\fR
+Shows cpufreq statistics if available.
+.TP  
+\fB\-y\fR \fB\-\-latency\fR
+Determines the maximum latency on CPU frequency changes.
+.TP  
+\fB\-o\fR \fB\-\-proc\fR
+Prints out information like provided by the /proc/cpufreq interface in 2.4. and early 2.6. kernels.
+.TP  
+\fB\-m\fR \fB\-\-human\fR
+human\-readable output for the \-f, \-w, \-s and \-y parameters.
+.TP  
+\fB\-h\fR \fB\-\-help\fR
+Prints out the help screen.
+.SH "REMARKS"
+.LP 
+By default only values of core zero are displayed. How to display settings of
+other cores is described in the cpupower(1) manpage in the \-\-cpu option section.
+.LP 
+You can't specify more than one of the output specific options \-o \-e \-a \-g \-p \-d \-l \-w \-f \-y.
+.LP 
+You also can't specify the \-o option combined with the \-c option.
+.SH "FILES"
+.nf 
+\fI/sys/devices/system/cpu/cpu*/cpufreq/\fP  
+\fI/proc/cpufreq\fP (deprecated) 
+\fI/proc/sys/cpu/\fP (deprecated)
+.fi 
+.SH "AUTHORS"
+.nf
+Dominik Brodowski <linux@brodo.de> \- author 
+Mattia Dongili<malattia@gmail.com> \- first autolibtoolization
+.fi
+.SH "SEE ALSO"
+.LP 
+cpupower\-frequency\-set(1), cpupower(1)

tools/power/cpupower/man/cpupower-frequency-set.1

+.TH "cpufreq-set" "1" "0.1" "Mattia Dongili" ""
+.SH "NAME"
+.LP 
+cpufreq\-set \- A small tool which allows to modify cpufreq settings.
+.SH "SYNTAX"
+.LP 
+cpufreq\-set [\fIoptions\fP]
+.SH "DESCRIPTION"
+.LP 
+cpufreq\-set allows you to modify cpufreq settings without having to type e.g. "/sys/devices/system/cpu/cpu0/cpufreq/scaling_set_speed" all the time.
+.SH "OPTIONS"
+.LP 
+.TP 
+\fB\-d\fR \fB\-\-min\fR <FREQ>
+new minimum CPU frequency the governor may select.
+.TP 
+\fB\-u\fR \fB\-\-max\fR <FREQ>
+new maximum CPU frequency the governor may select.
+.TP 
+\fB\-g\fR \fB\-\-governor\fR <GOV>
+new cpufreq governor.
+.TP 
+\fB\-f\fR \fB\-\-freq\fR <FREQ>
+specific frequency to be set. Requires userspace governor to be available and loaded.
+.TP 
+\fB\-r\fR \fB\-\-related\fR
+modify all hardware-related CPUs at the same time
+.TP 
+\fB\-h\fR \fB\-\-help\fR
+Prints out the help screen.
+.SH "REMARKS"
+.LP 
+By default values are applied on all cores. How to modify single core
+configurations is described in the cpupower(1) manpage in the \-\-cpu option section.
+.LP 
+The \-f FREQ, \-\-freq FREQ parameter cannot be combined with any other parameter.
+.LP 
+FREQuencies can be passed in Hz, kHz (default), MHz, GHz, or THz by postfixing the value with the wanted unit name, without any space (frequency in kHz =^ Hz * 0.001 =^ MHz * 1000 =^ GHz * 1000000).
+.LP 
+On Linux kernels up to 2.6.29, the \-r or \-\-related parameter is ignored.
+.SH "FILES" 
+.nf
+\fI/sys/devices/system/cpu/cpu*/cpufreq/\fP  
+\fI/proc/cpufreq\fP (deprecated) 
+\fI/proc/sys/cpu/\fP (deprecated)
+.fi 
+.SH "AUTHORS"
+.nf 
+Dominik Brodowski <linux@brodo.de> \- author 
+Mattia Dongili<malattia@gmail.com> \- first autolibtoolization
+.fi
+.SH "SEE ALSO"
+.LP 
+cpupower\-frequency\-info(1), cpupower(1)

tools/power/cpupower/man/cpupower-info.1

+.TH CPUPOWER\-INFO "1" "22/02/2011" "" "cpupower Manual"
+.SH NAME
+cpupower\-info \- Shows processor power related kernel or hardware configurations
+.SH SYNOPSIS
+.ft B
+.B cpupower info [ \-b ] [ \-s ] [ \-m ]
+
+.SH DESCRIPTION
+\fBcpupower info \fP shows kernel configurations or processor hardware
+registers affecting processor power saving policies.
+
+Some options are platform wide, some affect single cores. By default values
+of core zero are displayed only. cpupower --cpu all cpuinfo will show the
+settings of all cores, see cpupower(1) how to choose specific cores.
+
+.SH "SEE ALSO"
+Options are described in detail in:
+
+cpupower(1), cpupower-set(1)

tools/power/cpupower/man/cpupower-monitor.1

+.TH CPUPOWER\-MONITOR "1" "22/02/2011" "" "cpupower Manual"
+.SH NAME
+cpupower\-monitor \- Report processor frequency and idle statistics
+.SH SYNOPSIS
+.ft B
+.B cpupower monitor
+.RB "\-l"
+
+.B cpupower monitor
+.RB [ "\-m <mon1>," [ "<mon2>,..." ] ]
+.RB [ "\-i seconds" ]
+.br
+.B cpupower monitor
+.RB [ "\-m <mon1>," [ "<mon2>,..." ] ]
+.RB command
+.br
+.SH DESCRIPTION
+\fBcpupower-monitor \fP reports processor topology, frequency and idle power
+state statistics. Either \fBcommand\fP is forked and
+statistics are printed upon its completion, or statistics are printed periodically.
+
+\fBcpupower-monitor \fP implements independent processor sleep state and
+frequency counters. Some are retrieved from kernel statistics, some are
+directly reading out hardware registers. Use \-l to get an overview which are
+supported on your system.
+
+.SH Options
+.PP
+\-l
+.RS 4
+List available monitors on your system. Additional details about each monitor
+are shown:
+.RS 2
+.IP \(bu
+The name in quotation marks which can be passed to the \-m parameter.
+.IP \(bu
+The number of different counters the monitor supports in brackets.
+.IP \(bu
+The amount of time in seconds the counters might overflow, due to
+implementation constraints.
+.IP \(bu
+The name and a description of each counter and its processor hierarchy level
+coverage in square brackets:
+.RS 4
+.IP \(bu
+[T] \-> Thread
+.IP \(bu
+[C] \-> Core
+.IP \(bu
+[P] \-> Processor Package (Socket)
+.IP \(bu
+[M] \-> Machine/Platform wide counter
+.RE
+.RE
+.RE
+.PP
+\-m <mon1>,<mon2>,...
+.RS 4
+Only display specific monitors. Use the monitor string(s) provided by \-l option.
+.RE
+.PP
+\-i seconds
+.RS 4
+Measure intervall.
+.RE
+.PP
+command
+.RS 4
+Measure idle and frequency characteristics of an arbitrary command/workload.
+The executable \fBcommand\fP is forked and upon its exit, statistics gathered since it was
+forked are displayed.
+.RE
+.PP
+\-v
+.RS 4
+Increase verbosity if the binary was compiled with the DEBUG option set.
+.RE
+
+.SH MONITOR DESCRIPTIONS
+.SS "Idle_Stats"
+Shows statistics of the cpuidle kernel subsystem. Values are retrieved from
+/sys/devices/system/cpu/cpu*/cpuidle/state*/.
+The kernel updates these values every time an idle state is entered or
+left. Therefore there can be some inaccuracy when cores are in an idle
+state for some time when the measure starts or ends. In worst case it can happen
+that one core stayed in an idle state for the whole measure time and the idle
+state usage time as exported by the kernel did not get updated. In this case
+a state residency of 0 percent is shown while it was 100.
+
+.SS "Mperf"
+The name comes from the aperf/mperf (average and maximum) MSR registers used
+which are available on recent X86 processors. It shows the average frequency
+(including boost frequencies).
+The fact that on all recent hardware the mperf timer stops ticking in any idle
+state it is also used to show C0 (processor is active) and Cx (processor is in
+any sleep state) times. These counters do not have the inaccuracy restrictions
+the "Idle_Stats" counters may show.
+May work poorly on Linux-2.6.20 through 2.6.29, as the \fBacpi-cpufreq \fP
+kernel frequency driver periodically cleared aperf/mperf registers in those
+kernels.
+
+.SS "Nehalem" "SandyBridge"
+Intel Core and Package sleep state counters.
+Threads (hyperthreaded cores) may not be able to enter deeper core states if
+its sibling is utilized.
+Deepest package sleep states may in reality show up as machine/platform wide
+sleep states and can only be entered if all cores are idle. Look up Intel
+manuals (some are provided in the References section) for further details.
+
+.SS "Ontario" "Liano"
+AMD laptop and desktop processor (family 12h and 14h) sleep state counters.
+The registers are accessed via PCI and therefore can still be read out while
+cores have been offlined.
+
+There is one special counter: NBP1 (North Bridge P1).
+This one always returns 0 or 1, depending on whether the North Bridge P1
+power state got entered at least once during measure time.
+Being able to enter NBP1 state also depends on graphics power management.
+Therefore this counter can be used to verify whether the graphics' driver
+power management is working as expected.
+
+.SH EXAMPLES
+
+cpupower monitor -l" may show:
+.RS 4
+Monitor "Mperf" (3 states) \- Might overflow after 922000000 s
+
+   ...
+
+Monitor "Idle_Stats" (3 states) \- Might overflow after 4294967295 s
+
+   ...
+
+.RE
+cpupower monitor \-m "Idle_Stats,Mperf" scp /tmp/test /nfs/tmp
+
+Monitor the scp command, show both Mperf and Idle_Stats states counter
+statistics, but in exchanged order.
+
+
+
+.RE
+Be careful that the typical command to fully utilize one CPU by doing:
+
+cpupower monitor cat /dev/zero >/dev/null
+
+Does not work as expected, because the measured output is redirected to
+/dev/null. This could get workarounded by putting the line into an own, tiny
+shell script. Hit CTRL\-c to terminate the command and get the measure output
+displayed.
+
+.SH REFERENCES
+"BIOS and Kernel Developer’s Guide (BKDG) for AMD Family 14h Processors"
+http://support.amd.com/us/Processor_TechDocs/43170.pdf
+
+"Intel® Turbo Boost Technology
+in Intel® Core™ Microarchitecture (Nehalem) Based Processors"
+http://download.intel.com/design/processor/applnots/320354.pdf
+
+"Intel® 64 and IA-32 Architectures Software Developer's Manual
+Volume 3B: System Programming Guide"
+http://www.intel.com/products/processor/manuals
+
+.SH FILES
+.ta
+.nf
+/dev/cpu/*/msr
+/sys/devices/system/cpu/cpu*/cpuidle/state*/.
+.fi
+
+.SH "SEE ALSO"
+powertop(8), msr(4), vmstat(8)
+.PP
+.SH AUTHORS
+.nf
+Written by Thomas Renninger <trenn@suse.de>
+
+Nehalem, SandyBridge monitors and command passing
+based on turbostat.8 from Len Brown <len.brown@intel.com>

tools/power/cpupower/man/cpupower-set.1

+.TH CPUPOWER\-SET "1" "22/02/2011" "" "cpupower Manual"
+.SH NAME
+cpupower\-set \- Set processor power related kernel or hardware configurations
+.SH SYNOPSIS
+.ft B
+.B cpupower set [ \-b VAL ] [ \-s VAL ] [ \-m VAL ]
+
+
+.SH DESCRIPTION
+\fBcpupower set \fP sets kernel configurations or directly accesses hardware
+registers affecting processor power saving policies.
+
+Some options are platform wide, some affect single cores. By default values
+are applied on all cores. How to modify single core configurations is
+described in the cpupower(1) manpage in the \-\-cpu option section. Whether an
+option affects the whole system or can be applied to individual cores is
+described in the Options sections.
+
+Use \fBcpupower info \fP to read out current settings and whether they are
+supported on the system at all.
+
+.SH Options
+.PP
+\-\-perf-bias, \-b
+.RS 4
+Sets a register on supported Intel processore which allows software to convey
+its policy for the relative importance of performance versus energy savings to
+the  processor.
+
+The range of valid numbers is 0-15, where 0 is maximum
+performance and 15 is maximum energy efficiency.
+
+The processor uses this information in model-specific ways
+when it must select trade-offs between performance and
+energy efficiency.
+
+This policy hint does not supersede Processor Performance states
+(P-states) or CPU Idle power states (C-states), but allows
+software to have influence where it would otherwise be unable
+to express a preference.
+
+For example, this setting may tell the hardware how
+aggressively or conservatively to control frequency
+in the "turbo range" above the explicitly OS-controlled
+P-state frequency range.  It may also tell the hardware
+how aggressively it should enter the OS requested C-states.
+
+This option can be applied to individual cores only via the \-\-cpu option,
+cpupower(1).
+
+Setting the performance bias value on one CPU can modify the setting on
+related CPUs as well (for example all CPUs on one socket), because of
+hardware restrictions.
+Use \fBcpupower -c all info -b\fP to verify.
+
+This options needs the msr kernel driver (CONFIG_X86_MSR) loaded.
+.RE
+.PP
+\-\-sched\-mc,  \-m [ VAL ]
+.RE
+\-\-sched\-smt, \-s [ VAL ]
+.RS 4
+\-\-sched\-mc utilizes cores in one processor package/socket first before
+processes are scheduled to other processor packages/sockets.
+
+\-\-sched\-smt utilizes thread siblings of one processor core first before
+processes are scheduled to other cores.
+
+The impact on power consumption and performance (positiv or negativ) heavily
+depends on processor support for deep sleep states, frequency scaling and
+frequency boost modes and their dependencies between other thread siblings
+and processor cores.
+
+Taken over from kernel documentation:
+
+Adjust the kernel's multi-core scheduler support.
+
+Possible values are:
+.RS 2
+0 - No power saving load balance (default value)
+
+1 - Fill one thread/core/package first for long running threads
+
+2 - Also bias task wakeups to semi-idle cpu package for power
+savings
+.RE
+
+sched_mc_power_savings is dependent upon SCHED_MC, which is
+itself architecture dependent.
+
+sched_smt_power_savings is dependent upon SCHED_SMT, which
+is itself architecture dependent.
+
+The two files are independent of each other. It is possible
+that one file may be present without the other.
+
+.SH "SEE ALSO"
+cpupower-info(1), cpupower-monitor(1), powertop(1)
+.PP
+.SH AUTHORS
+.nf
+\-\-perf\-bias parts written by Len Brown <len.brown@intel.com>
+Thomas Renninger <trenn@suse.de>

tools/power/cpupower/man/cpupower.1

+.TH CPUPOWER "1" "07/03/2011" "" "cpupower Manual"
+.SH NAME
+cpupower \- Shows and sets processor power related values
+.SH SYNOPSIS
+.ft B
+.B cpupower [ \-c cpulist ] subcommand [ARGS]
+
+.B cpupower \-v|\-\-version
+
+.B cpupower \-h|\-\-help
+
+.SH DESCRIPTION
+\fBcpupower \fP is a collection of tools to examine and tune power saving
+related features of your processor.
+
+The manpages of the subcommands (cpupower\-<subcommand>(1)) provide detailed
+descriptions of supported features. Run \fBcpupower help\fP to get an overview
+of supported subcommands.
+
+.SH Options
+.PP
+\-\-help, \-h
+.RS 4
+Shows supported subcommands and general usage.
+.RE
+.PP
+\-\-cpu cpulist,  \-c cpulist
+.RS 4
+Only show or set values for specific cores.
+This option is not supported by all subcommands, details can be found in the
+manpages of the subcommands.
+
+Some subcommands access all cores (typically the *\-set commands), some only
+the first core (typically the *\-info commands) by default.
+
+The syntax for <cpulist> is based on how the kernel exports CPU bitmasks via
+sysfs files. Some examples:
+.RS 4
+.TP 16
+Input
+Equivalent to
+.TP
+all
+all cores
+.TP
+0\-3
+0,1,2,3
+.TP
+0\-7:2
+0,2,4,6
+.TP
+1,3,5-7
+1,3,5,6,7
+.TP
+0\-3:2,8\-15:4
+0,2,8,12	
+.RE
+.RE
+.PP
+\-\-version,  \-v
+.RS 4
+Print the package name and version number.
+
+.SH "SEE ALSO"
+cpupower-set(1), cpupower-info(1), cpupower-idle(1),
+cpupower-frequency-set(1), cpupower-frequency-info(1), cpupower-monitor(1),
+powertop(1)
+.PP
+.SH AUTHORS
+.nf
+\-\-perf\-bias parts written by Len Brown <len.brown@intel.com>
+Thomas Renninger <trenn@suse.de>

tools/power/cpupower/utils/builtin.h

+#ifndef BUILTIN_H
+#define BUILTIN_H
+
+extern int cmd_set(int argc, const char **argv);
+extern int cmd_info(int argc, const char **argv);
+extern int cmd_freq_set(int argc, const char **argv);
+extern int cmd_freq_info(int argc, const char **argv);
+extern int cmd_idle_info(int argc, const char **argv);
+extern int cmd_monitor(int argc, const char **argv);
+
+extern void set_help(void);
+extern void info_help(void);
+extern void freq_set_help(void);
+extern void freq_info_help(void);
+extern void idle_info_help(void);
+extern void monitor_help(void);
+
+#endif