twister: using unsupported fixture without defined harness causes an infinite loop during on-target test execution

[PerMac]

Describe the bug
The bug can be observed when the following conditions combine:

• a test requiring a fixture
• the test does not have harness: console or ztest
• the test is called for hardware that does not have the fixture defined.

Even though a test requires a fixture in its yaml, twister will get in an infinite loop waiting for the device to become ready. I think the logic in handle() (line 689 in twisterlib.py) is wrong:

        while not self.device_is_available(self.instance):
            logger.debug("Waiting for device {} to become available".format(self.instance.platform.name))
            time.sleep(1)

The function device_is_available() checks if the fixture exists in the device and returns False. Then the 1 sec sleep is applied and the check for the fixture repeats. This happens until the timeout termination.
If the test defines harness as console or ztest then the test is skipped correctly. If none of these harnesses is used then we end up in the loop.

I am not sure if it is a real error. I found it when I was trying to figure out why #31188 was not getting filtered (now harness: ztest is added to the yamls in PR and fixture is skipped correctly).

[PerMac]

@nashif related question: Why in testcase_runnable() (line 1761 of twisterlib.py) fixture requirement is checked only if testcase.harness in [ 'console', 'ztest']?

[nashif]

@nashif related question: Why in testcase_runnable() (line 1761 of twisterlib.py) fixture requirement is checked only if testcase.harness in [ 'console', 'ztest']?

because is the only way we have to deal with test output from fixtures

[nashif]

a harness_config containing a fixture always needs a fixture in the yaml file. So this is more of a schema dependency validation issue and less of a logical issue I think.

[nashif]

dont think this is a bug, it is just bad test configuration.