Image tests

About the the testimage class

The build system has the ability to run a series of automated tests for qemu images.

All the tests are actually commands run on the target system over ssh.

The tests themselves are written in Python, making use of the unittest module.

The class that enables this is testimage.bbclass (which handles loading the tests and starting the qemu image)

Enabling and running the tests

Requirements

You should be aware of the following:

• runqemu script needs sudo access for setting up the tap interface, so you need to make sure it can do that non-interactively. That means you need to do one of the following:
  • add NOPASSWD for your user either for ALL commands in /etc/sudoers, either just for runqemu-ifup (but you need to provide the full path and that can change if you have multiple poky clones)
    • on some distributions you also need to comment out "Defaults requiretty" in /etc/sudoers
  • manually configure a tap interface for your system
  • run as root the script in scripts/runqemu-gen-tapdev which should generate a list of tap devices (that's usually done in AutoBuilder-like setups)
• the DISPLAY variable needs to be set so that means you need to have an X server available (e.g start vncserver)
• some of the tests (in particular smart tests) start a http server on a random high number port, used to serve files to the target. The smart module for example serves ${DEPLOY_DIR}/rpm to the target so we run smart channel commands. That means your host's firewall should accept incoming connection from 192.168.7.0/24 (the default class used for tap0 devices by runqemu)

Known bugs/limitations

• for builds where rm_work was used most of the tests will error out because they rely on ${WORKDIR}/installed_pkgs.txt (see https://bugzilla.yoctoproject.org/show_bug.cgi?id=5072 )

Usage

To use it add "testimage" to global inherit and call your target image with -c testimage, like this:

• for example build a qemu core-image-sato: bitbake core-image-sato
• add INHERIT += "testimage" in local.conf
• then call "bitbake core-image-sato -c testimage". That will run a standard suite of tests.

The name of tests is the name of the python modules in meta/lib/oeqa/runtime.

A layer can add its own tests in <meta-layer>/lib/oeqa/runtime, provided it extends BBPATH as normal in its layer.conf (test module names shouldn't collide though with those in core).

You can change the tests run by appending or overrding the TEST_SUITES variable in local.conf. Each name in TEST_SUITES represents a required test for the image. That means that no skipping is allowed (even if the test isn't suitable for the image, e.g running the rpm tests on a images with no rpm). Appending "auto" to TEST_SUITES means that it will try to run all tests that are suitable for the image (each test decides that on it's own).

Note that the order in TEST_SUITES is important (it's the order modules run) and it influences tests dependencies. That means that tests that depend on other tests (e.g ssh depends on the ping test) should be added last. Each module can have multiple classes with multiple test methods (and Python unittest rules apply here).

Examples

Examples:

• to run the default tests for core-image-sato you don't need to change TEST_SUITES.
• The default for core-image-sato is defined as: DEFAULT_TEST_SUITES_pn-core-image-sato = "ping ssh df connman syslog xorg scp vnc date rpm smart dmesg"
• to add your own test to the list of the defaults add: TEST_SUITES_append = " mytest"
• to run a specific list of tests: TEST_SUITES = "ping ssh rpm" (remember order is important)

Once you call the testimage task (bitbake <my-image> -c testimage) a couple of things happen:

• a copy of the rootfs is done in ${WORKDIR}/testimage
• the image is booted under qemu using the standard runqemu script
• there is a timeout of 500 seconds by default for the boot process to reach the login prompt (you can change the timeout by setting TEST_QEMUBOOT_TIMEOUT in local.conf)
• once the boot process reached the login prompt the tests are run (you can find the full boot log in ${WORKDIR}/testimage/qemu_boot_log)
• each test module is loaded in the order found in TEST_SUITES (the full output of the commands ran over ssh is found in ${WORKDIR}/testimgage/ssh_target_log)
• if there are no fails, the task will end successfully. You can find the output from the unittest in the task log (in ${WORKDIR}/temp/log.do_testimage)

---

Log for a custom, systemd-enabled image that has package-management feature and TEST_SUITES = "ping ssh rpm auto" in local.conf

$ cat tmp/work/qemux86_64-poky-linux/core-image-base/1.0-r0/temp/log.do_testimage
DEBUG: Executing python function do_testimage
NOTE: Created listening socket for qemu serial console on: 127.0.0.1:56358
NOTE: DISPLAY value: :0
NOTE: rootfs file: /home/stefans/yocto/builds/firefly/tmp/work/qemux86_64-poky-linux/core-image-base/1.0-r0/testimage/core-image-base-qemux86-64-testimage.ext3
NOTE: Qemu log file: /home/stefans/yocto/builds/firefly/tmp/work/qemux86_64-poky-linux/core-image-base/1.0-r0/testimage/qemu_boot_log.20130819115123
NOTE: SSH log file: /home/stefans/yocto/builds/firefly/tmp/work/qemux86_64-poky-linux/core-image-base/1.0-r0/testimage/ssh_target_log.20130819115123
NOTE: runqemu started, pid is 2979
NOTE: waiting at most 60 seconds for qemu pid
NOTE: qemu started - qemu procces pid is 3061
NOTE: IP found: 192.168.7.2
NOTE: Waiting at most 500 seconds for login banner
NOTE: Connection from 127.0.0.1:44406
NOTE: Reached login banner
NOTE: Test modules  ['oeqa.runtime.ping', 'oeqa.runtime.ssh', 'oeqa.runtime.rpm', 'oeqa.runtime.multilib', 'oeqa.runtime.smart', 'oeqa.runtime.dmesg', 'oeqa.runtime.df', 'oeqa.runtime.connman', 'oeqa.runtime.gcc', 'oeqa.runtime.xorg', 'oeqa.runtime.syslog', 'oeqa.runtime.systemd']
NOTE: Found 31 tests
test_ping (oeqa.runtime.ping.PingTest) ... ok
test_ssh (oeqa.runtime.ssh.SshTest) ... ok
test_rpm_help (oeqa.runtime.rpm.RpmHelpTest) ... ok
test_rpm_query (oeqa.runtime.rpm.RpmQueryTest) ... ok
skipped "multilib: this isn't a multilib:lib32 image"
test_smart_help (oeqa.runtime.smart.SmartHelpTest) ... ok
test_smart_info (oeqa.runtime.smart.SmartQueryTest) ... ok
test_smart_query (oeqa.runtime.smart.SmartQueryTest) ... ok
test_dmesg (oeqa.runtime.dmesg.DmesgTest) ... ok
test_df (oeqa.runtime.df.DfTest) ... ok
skipped 'connman: No connman package in image'
skipped "gcc: Image doesn't have tools-sdk in IMAGE_FEATURES"
skipped "xorg: target doesn't have x11 in IMAGE_FEATURES"
test_syslog_help (oeqa.runtime.syslog.SyslogTest) ... ok
test_syslog_running (oeqa.runtime.syslog.SyslogTest) ... ok
test_syslog_logger (oeqa.runtime.syslog.SyslogTestConfig) ... ok
test_syslog_restart (oeqa.runtime.syslog.SyslogTestConfig) ... ok
test_syslog_startup_config (oeqa.runtime.syslog.SyslogTestConfig) ... skipped 'Not appropiate for systemd image'
test_systemd_version (oeqa.runtime.systemd.SystemdBasicTest) ... ok
test_systemd_disable (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_enable (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_failed (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_list (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_service (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_start (oeqa.runtime.systemd.SystemdTests) ... ok
test_systemd_stop (oeqa.runtime.systemd.SystemdTests) ... ok

----------------------------------------------------------------------
Ran 22 tests in 48.492s

OK (skipped=5)
NOTE: All required tests passed
DEBUG: Python function do_testimage finished

As you can see some tests passed and some of them were skipped (because they weren't applicable for this image). And while I haven't added systemd tests to TEST_SUITES the tests were run (because of auto).

---

Let's see what happens if I use TEST_SUITES = "ping ssh gcc" for a core-image-sato image (which doesn't have the tools-sdk feature):

--snip--
NOTE: Reached login banner
NOTE: Test modules  ['oeqa.runtime.ping', 'oeqa.runtime.ssh', 'oeqa.runtime.gcc']
NOTE: Found 5 tests
test_ping (oeqa.runtime.ping.PingTest) ... ok
test_ssh (oeqa.runtime.ssh.SshTest) ... ok
ERROR

======================================================================
ERROR: setUpModule (oeqa.runtime.gcc)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/stefans/z/poky/meta/lib/oeqa/runtime/gcc.py", line 8, in setUpModule
    skipModule("Image doesn't have tools-sdk in IMAGE_FEATURES")
  File "/home/stefans/z/poky/meta/lib/oeqa/oetest.py", line 108, in skipModule
    "\nor the image really doesn't have the requred feature/package when it should." % (modname, reason))
Exception: 
Test gcc wants to be skipped.
Reason is: Image doesn't have tools-sdk in IMAGE_FEATURES
Test was required in TEST_SUITES, so either the condition for skipping is wrong
or the image really doesn't have the requred feature/package when it should.

----------------------------------------------------------------------
Ran 2 tests in 6.255s

FAILED (errors=1)
NOTE: Sending SIGTERM to runqemu
DEBUG: Python function do_testimage finished
ERROR: Function failed: Some tests failed. You should check the task log and the ssh log. (ssh log is /home/stefans/z/poky/build/tmp/work/qemux86_64-poky-linux/core-image-sato/1.0-r0/testimage/ssh_target_log.20130827122341

• First, it tells us it loaded the module we required (ping, ssh and gcc) and that there are 5 tests (because the gcc module has 3 test methods)
• It starts running the tests
• the gcc module will error out giving us a traceback of why that happened. Because gcc was a required test, it wasn't skipped like earlier, instead it was marked as an error.

---

Some random examples from the ssh log (tests run here are ssh, df, syslog, xorg, rpm, dmesg)

Good to know:

• Q: why is there a . /etc/profile before each command? A: Because of the default PATH (/bin:/usr/bin) when running commands over ssh (the answer is a bit more complex, let's just say we need to source /etc/profile for extending PATH)
• while it might look that the commands aren't properly escaped those ssh commands are actually run through Python's subprocess module with shell=False (so copy-paste of the commands in your shell won't work unless you properly escape them)
• there is a default timeout of 300 seconds for each command (though a test can overwrite that or run a command with no timeout). There is no timeout for scp commands though.
• the tests can use the return code and/or the output to decide if they fail/pass.
• in almost all cases, the first line of the output is stripped (which is: Warning: Permanently added '192.168.7.2' (RSA) to the list of known hosts.)

[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; uname -a
Linux qemux86-64 3.8.13-yocto-standard #2 SMP PREEMPT Mon Aug 26 17:10:37 EEST 2013 x86_64 GNU/Linux
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; which df
/bin/df
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; df / | sed -n '2p' | awk '{print $4}'
110560
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; logger foobar && grep foobar /var/log/messages
Aug 27 12:34:57 qemux86-64 user.notice root: foobar
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; /etc/init.d/syslog restart
Stopping syslogd/klogd: stopped syslogd (pid 699)
stopped klogd (pid 709)
done
Starting syslogd/klogd: done
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; echo "LOGFILE=/var/log/test" >> /etc/syslog-startup.conf

[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; /etc/init.d/syslog restart
Stopping syslogd/klogd: stopped syslogd (pid 808)
stopped klogd (pid 811)
done
Starting syslogd/klogd: done
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; logger foobar && grep foobar /var/log/test
Aug 27 12:35:03 qemux86-64 user.notice root: foobar
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; sed -i 's#LOGFILE=/var/log/test##' /etc/syslog-startup.conf

[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; /etc/init.d/syslog restart
Stopping syslogd/klogd: stopped syslogd (pid 824)
stopped klogd (pid 826)
done
Starting syslogd/klogd: done
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; cat /var/log/Xorg.0.log | grep -v "(EE) error," | grep -v "PreInit" | grep -v "evdev:" | grep -v "glx" | grep "(EE)"

[SSH command returned]: 1
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; ps |  grep -v xinit | grep [X]org
  601 root     79776 S <  /usr/bin/Xorg :0 -br -pn
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; rpm -q rpm
rpm-5.4.9-r63.x86_64
[SSH command returned]: 0
[Running SCP]$ scp -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no /home/stefans/z/poky/build/tmp/deploy/rpm/x86_64/rpm-doc-5.4.9-r63.x86_64.rpm root@192.168.7.2:/tmp/rpm-doc.rpm
Warning: Permanently added '192.168.7.2' (RSA) to the list of known hosts.

[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; rpm -ivh /tmp/rpm-doc.rpm
Preparing...                ##################################################
rpm-doc                     ##################################################
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; rpm -e rpm-doc

[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; rm -f /tmp/rpm-doc.rpm

[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; which dmesg
/bin/dmesg
[SSH command returned]: 0
[Running]$ ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l root 192.168.7.2 . /etc/profile; dmesg | grep -v mmci-pl18x | grep -v "error changing net interface name" | grep -i error

[SSH command returned]: 1

Writing new tests

All new test modules should go in meta/lib/oeqa/runtime (or <meta-layer>/lib/oeqa/runtime if that test is meant for a certain layer only). You could start by copying an existing module (syslog.py or gcc.py) are good examples and go from there.