check for the timer interrupt raised on cpu for duration of
3 seconds and report the mode
Signed-off-by: Sanjay Singh Rawat <sanjay.rawat(a)linaro.org>
---
Makefile | 2 ++
nohz/Makefile | 25 +++++++++++++++++++++
nohz/nohz_01.sh | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
nohz/nohz_01.txt | 1 +
4 files changed, 92 insertions(+)
create mode 100644 nohz/Makefile
create mode 100755 nohz/nohz_01.sh
create mode 100644 nohz/nohz_01.txt
diff --git a/Makefile b/Makefile
index 731619d..896497c 100644
--- a/Makefile
+++ b/Makefile
@@ -34,6 +34,7 @@ check:
@(cd thermal; $(MAKE) check)
# @(cd powertop; $(MAKE) check)
@(cd cputopology; $(MAKE) check)
+ @(cd nohz; $(MAKE) check)
uncheck:
@(cd cpufreq; $(MAKE) uncheck)
@@ -41,6 +42,7 @@ uncheck:
@(cd cpuidle; $(MAKE) uncheck)
# @(cd suspend; $(MAKE) uncheck)
@(cd thermal; $(MAKE) uncheck)
+ @(cd nohz; $(MAKE) uncheck)
recheck: uncheck check
diff --git a/nohz/Makefile b/nohz/Makefile
new file mode 100644
index 0000000..315ff6f
--- /dev/null
+++ b/nohz/Makefile
@@ -0,0 +1,25 @@
+#
+# PM-QA validation test suite for the power management on Linux
+#
+# Copyright (C) 2014, Linaro Limited.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation; either version 2
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+#
+# Contributors:
+# Sanjay Singh Rawat <sanjay.rawat(a)linaro.org> (LG Electronics)
+# - initial API and implementation
+#
+
+include ../Test.mk
diff --git a/nohz/nohz_01.sh b/nohz/nohz_01.sh
new file mode 100755
index 0000000..2828188
--- /dev/null
+++ b/nohz/nohz_01.sh
@@ -0,0 +1,64 @@
+#!/bin/bash
+#
+# PM-QA validation test suite for the power management on Linux
+#
+# Copyright (C) 2014, Linaro Limited.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation; either version 2
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+#
+# Contributors:
+# Sanjay Singh Rawat <sanjay.rawat(a)linaro.org> (LG Electronics)
+# - initial API and implementation
+#
+
+# URL : https://wiki.linaro.org/WorkingGroups/PowerManagement/Resources/TestSuite/P…
+
+source ../include/functions.sh
+
+check_cpu_nohz_interrupt() {
+ local cpu=$1
+ local cpuid=${cpu:3}
+ local ints=
+ local inte=
+ local i=
+ local diff=
+
+ i=$(cat /proc/interrupts | grep time | wc -l)
+ if [ $i -ne 1 ]; then
+ echo "warning: multiple entry for timer interrupt :" $i
+ exit 1
+ fi
+
+ if [ "$cpu" == "cpu0" ]; then
+ return 0
+ fi
+
+ cpuid=$((cpuid+2))
+ ints=$(cat /proc/interrupts | grep time | awk -v c=$cpuid '{ print $c }')
+ sleep 3
+ inte=$(cat /proc/interrupts | grep time | awk -v c=$cpuid '{ print $c }')
+
+ #echo $cpu "interrupts before" $ints "after" $inte
+ diff=$((inte-ints))
+
+ if [ $diff -eq 0 ]; then
+ echo $cpu "in NOHZ"
+ else
+ echo $cpu "not in NOHZ"
+ fi
+}
+
+for_each_cpu check_cpu_nohz_interrupt
+test_status_show
diff --git a/nohz/nohz_01.txt b/nohz/nohz_01.txt
new file mode 100644
index 0000000..0c547f5
--- /dev/null
+++ b/nohz/nohz_01.txt
@@ -0,0 +1 @@
+check each cpu is in nohz mode (wait 3 seconds for timer interrupt)
--
1.7.10.4
From: Amit Kucheria <amit.kucheria(a)linaro.org>
More cleanups to the command-line options to make idlestat easier to use and
separate the two modes: tracing and reporting.
Regards,
Amit
Amit Kucheria (6):
Use the correct version in the log file
Print more details when fopen fails
kick the dump option off to use -z and release -m for later
-o stands for output file when idlestat can take an input file as well
Introduce --trace and --import mode
Lots of changes related to usage, bump the version to 0.3
README | 10 ++++-----
idlestat.c | 71 +++++++++++++++++++++++++++++++++++++++++++-------------------
2 files changed, 54 insertions(+), 27 deletions(-)
--
1.9.1
Hello,
this email is to announce that ci.linaro.org will not be available on
Friday 6th June due to planned maintenance.
We will perform a server upgrade to the latest Ubuntu LTS release.
In order to perform a smooth upgrade, please make any update to your
jobs before Thursday 5th June at 12:00UTC.
Another email will be sent out when the service has been restored.
Regards.
--
Milo Casagrande
Linaro.org <www.linaro.org> │ Open source software for ARM SoCs
Hello,
I noticed some odd behaviors with idlestate.c/execute() function.
1. I wrote a sample program hello.c with infinite loop -
int main() {
while(1);
}
#sudo ../idlestat/idlestat -o /tmp/myoutput ../temp/hello
idlestat never terminates, because there is no alarm set, hello
program never terminates.
2. If I change and invoke idlestat with below command -
#sudo ../idlestat/idlestat -o /tmp/myoutput -t 5 ../temp/hello
idlestat outputs only below line and exits -
Total trace buffer: 153896 kB
What is happening is the below check fails -
if (WIFSIGNALED(status) && WTERMSIG(status) == SIGTERM) {
return 0;
}
here condition is always false. Because the status variable was
fetched before killing the process. So status does not contain valid signal
number.
if (waitpid(pid, &status, 0) < 0) {
if (errno != EINTR || !sigalrm)
goto again;
kill(pid, SIGTERM);
}
Possible Fix: call waitpid again after kill and fetch new status.
3. The below check makes idlestat dependent on the hello program run -
if (WIFEXITED(status) && !WEXITSTATUS(status)) {
}
If my program is written to return a non-zero value, the second
condition will be always false.
So with the below program -
int main() {
return 1;
}
idlestat outputs only since WEXITSTATUS(status) returns nonzero value -
Total trace buffer: 153896 kB
Possible Fix: Just check WIFEXITED(status) and ommit the fetching of
exit status.
In any case, it should not be allowed to run idlestat with "-t 0", since
that will results in problem 1 again.
--
Thanks,
- Meraj
Hello,
There is an issue related to perf which I am facing since 15 days. Hoping that the great minds here will help me to solve this.
I have a requirement to make perf tool work on a device having ARM architecture. But, on recording the tracepoint events and then running ./perf report, it shows the shared objects name as [unknown] and Symbols as 00000, whereas for software and hardware events I do not experience this issue.I have cross compiled the perf tool available on mailine and ported it on a device having ARM architecture.
Output on ARM based device with kernel 3.4.
./perf record -e kmem:kmalloc cal
[ perf record: Woken up 1 times to write data ][ perf record: Captured and wrote 0.007 MB perf.data (~321 samples) ]
./perf report
Overhead Command Shared Object Symbol
........................... ................................ ...........................................40.78% cal [unknown] [.]0000000031.6% cal [unknown] [.]00000000
[...]
On ubuntu 12.04, system (kernel 3.10) it works perfectly fine.
Output on x86 architecture
./perf record -e kmem:kmalloc gcalctool
[ perf record: Woken up 3 times to write data ][ perf record: Captured and wrote 0.27 MB perf.data (~845 samples) ]
./perf report --stdio
Overhead Command Shared Object Symbol
........................... ................................ ...........................................96.55% cal [kernel.kallsyms] [k] kmem_cache_alloc_trace3.45% cal [kernel.kallsyms] [.]__kmalloc
[...]
Keenly, awaiting for you help.
Thanks.
Sneha.
Hi
I was trying to run idlestat and with the below command output is -
# sudo ./idlestat -o /tmp/myoutput
fopen: No such file or directory
I looked into the source and it looks like its not entering in the if
condition in main -
1212 /* Acquisition time specified means we will get the traces */
1213 if (options.duration || args < argc) {
1214
1215 /* Read cpu topology info from sysfs */
1216 read_sysfs_cpu_topo();
...
1260 if (idlestat_store(options.filename))
1261 return -1;
}
Since duration argument is not given and args is actually equal to argc, if
condition is evaluated false.
So idlestat_store() is not creating the file "/tmp/myoutput" or storing any
info there.
As a result, call to idlestat_load is issues above error message.
1264 /* Load the idle states information */
1265 datas = idlestat_load(options.filename);
1266 if (!datas)
1267 return 1;
I think the fix is simply to change the if condition above and make it -
1212 /* Acquisition time specified means we will get the traces */
1213 if (options.duration || args <= argc) {
1214
}
--
Thanks,
-Meraj
Hi,
I am looking at in cpufreq.c
__cpufreq_governor(policy, CPUFREQ_GOV_STOP);
up_write(&policy->rwsem);
__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
down_write(&policy->rwsem);
Why do we need to unlock the lock before calling CPUFREQ_GOV_POLICY_EXIT.
Wont this result in a race condition when 2 cpus are calling policy
exit, one after another during governor switch?
--
Regards,
Sriram
Hi,
These are a few cleanups around idlestat commandline options while I was
working on reformating the report.
Regards,
Amit
Amit Kucheria (9):
'verbose' option isn't used anywhere, remove it
Sort options in alphabetic order of the single-char options
Adding missing 'duration' option in long_options
Specify which options need arguments and which don't to help
getopt_long()
Stick to -d for debug according to widely used convention
Make getopt_long handling more robust
struct option long_options doesn't need to be global
improve help - show all available options
idledebug? When did we ever call it that?
README | 7 +++++--
idlestat.c | 71 +++++++++++++++++++++++++++++++++++---------------------------
2 files changed, 45 insertions(+), 33 deletions(-)
--
1.9.1
