0day is a performance and regression test system developed by
Fengguan Wu of Intel's open source development center.

 * https://01.org/lkp/documentation/0-day-test-service
 * https://github.com/fengguang/lkp-tests.git


Puts each test (named a 'job') into a yaml file

= command line =
lkp is a command line tool for executing a test

some command line options are:
 * lkp install <test_package>
   * ls $LKP_SRC/jobs to see available jobs
 * lkp split-job <test_package>
 * lkp run
 * lkp result <testname>

= test yaml file =
Here's what a test looks like:

From the file tbench.yaml
{{{#!YellowBox
suite: tbench
testcase: tbench
category: benchmark

# upto 40% CPU cycles may be used by latency stats
disable_latency_stats: 1

nr_threads: 100%

cluster: cs-localhost

if role server:
  tbench-server:

if role client:
  tbench:
}}}

In the 'pkg' directory are some PKGBUILD files.  This appears to be
the package file format for Arch Linux. see https://wiki.archlinux.org/index.php/PKGBUILD


Each test has a 'maker script' that tells how to build and install the 
test program.   See pack/ebizzy for an example.


= directory structure =
Here is the directory structure for lkp-tests:
 * allot
 * bin
 * cluster
 * daemon
 * distro
 * doc
 * etc
 * filters
 * hosts
 * include
 * '''jobs''' - has parameters for a particular execution of the test program (similar to fuegos specs and plans) 
 * lib
 * lkp-exec
 * monitors
 * '''pack''' - has the maker script which does the fetch, build, install for a test program
 * params
 * pkg
 * plot
 * repo
 * rootfs
 * sbin
 * setup
 * spec
 * '''stats''' - has the results parser for each test - output is JSON
 * '''tests''' - has the main test script for each test
 * tmp
 * tools

= dependency processing =
 * in lkp-tests/tests/pack-dep - check_shared_package

 * in jobs/ftrace_onoff.yaml:need_memory: 2G
 * in jobs/phoronix-test-suite-needx.yaml:need_x true
 * in include/ndctl:need_kernel_headers: true
 * in include/ltp:need_kconfig: CONFIG_BLK_DEV_LOOP
 * in filters/need_cpu: if (( need_cpu > nr_cpu )); then
 * in jobs/README.md: need_*

The file include/kernel_selftests has the following content:
{{{#!YellowBox
need_kernel_headers: true
need_kconfig:
- CONFIG_TEST_FIRMWARE
- CONFIG_TEST_BPF
- CONFIG_TEST_USER_COPY
- CONFIG_MEMORY_NOTIFIER_ERROR_INJECT
- CONFIG_MEMORY_HOTPLUG_SPARSE=y
- CONFIG_NOTIFIER_ERROR_INJECTION
- CONFIG_FTRACE=y
}}}

= test execution =
job.yaml files are converted into job.sh files for execution by sbin/job2sh.

Items in the yaml file are converted either to environment variables, or
to references to setup scripts or test scripts.

See jobs/README.md for details.

= monitors =
Monitors are located in <top_dir>/monitors.

Each one is a shell script.  Many are wrappers around existing tools, the
output of which is used for the monitor output.

They often use a library <topdir>/lib/wait.sh.

When executed with no interval, they take a snapshot before and after the
test.

When executed with an interval, they take a snapshot at the specified interval.

Here's an example, from cpuidle:
{{{#!YellowBox
#!/bin/sh
# - interval

[ -f /sys/devices/system/cpu/cpu0/cpuidle/state0/name ] || exit 0

take_snapshot()
{
        echo time: $(date +%s.%N)

        for dir in /sys/devices/system/cpu/cpu*/cpuidle/state*/
        do
                cpu=${dir#/sys/devices/system/cpu/}
                cpu=${cpu%%/*}
                read state_name  < $dir/name
                read state_time  < $dir/time
                read state_usage < $dir/usage
                echo $cpu.${state_name}.time: ${state_time}
                echo $cpu.${state_name}.usage: ${state_usage}
        done
}

file=$(basename $0)

. $LKP_SRC/lib/wait.sh
setup_wait

if [ -z "$interval" ]; then
        take_snapshot
        wait_post_test
        take_snapshot
        exit
fi

while :
do
        take_snapshot
        wait_timeout $interval
done
}}}