Re: [PATCH v2 04/11] tests/acceptance: add kernel record/replay test for x86_64

[Philippe Mathieu-Daudé]

On 5/28/20 1:07 PM, Alex Bennée wrote:
> 
> Philippe Mathieu-Daudé <philmd@redhat.com> writes:
> 
>> On 5/28/20 9:12 AM, Pavel Dovgalyuk wrote:
>>>
>>> On 27.05.2020 17:53, Philippe Mathieu-Daudé wrote:
>>>> On 5/27/20 12:31 PM, Pavel Dovgalyuk wrote:
>>>>> This patch adds a test for record/replay an execution of x86_64 machine.
>>>>> Execution scenario includes simple kernel boot, which allows testing
>>>>> basic hardware interaction in RR mode.
>>>>>
>>>>> Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru>
>>>>> ---
>>>>>   0 files changed
>>>>>
>>>>> diff --git a/tests/acceptance/replay_kernel.py
>>>>> b/tests/acceptance/replay_kernel.py
>>>>> index b8b277ad2f..c7526f1aba 100644
>>>>> --- a/tests/acceptance/replay_kernel.py
>>>>> +++ b/tests/acceptance/replay_kernel.py
>>>>> @@ -55,3 +55,19 @@ class ReplayKernel(LinuxKernelUtils):
>>>>>                       True, shift, args)
>>>>>           self.run_vm(kernel_path, kernel_command_line, console_pattern,
>>>>>                       False, shift, args)
>>>>> +
>>>>> +    def test_x86_64_pc(self):
>>>>> +        """
>>>>> +        :avocado: tags=arch:x86_64
>>>>> +        :avocado: tags=machine:pc
>>>>> +        """
>>>>> +        kernel_url =
>>>>> ('https://archives.fedoraproject.org/pub/archive/fedora'
>>>>> +                     
>>>>> '/linux/releases/29/Everything/x86_64/os/images/pxeboot'
>>>>> +                      '/vmlinuz')
>>>>> +        kernel_hash = '23bebd2680757891cf7adedb033532163a792495'
>>>>> +        kernel_path = self.fetch_asset(kernel_url,
>>>>> asset_hash=kernel_hash)
>>>>> +
>>>>> +        kernel_command_line = self.KERNEL_COMMON_COMMAND_LINE +
>>>>> 'console=ttyS0'
>>>>> +        console_pattern = 'Kernel command line: %s' %
>>>>> kernel_command_line
>>>>> +
>>>>> +        self.run_rr(kernel_path, kernel_command_line, console_pattern)
>>>>>
>>>> This one timeouted (I build with --enable-debug):
>>>
>>> I've got the strange behavior for the couple of times.
>>>
>>> Console output was correct (I saw 'Kernel command line' in logs), but
>>> _console_interation function didn't notice it.
>>>
>>> Therefore the test finished with timeout.
>>>
>>> How this could be possible?
>>
>> IIRC there is a problem in how Avocado consume the chardev output.
> 
> Is this the same problem Robert has tried to work around in tests/vm?
> 
>   From: Robert Foley <robert.foley@linaro.org>
>   Subject: [PATCH v7 12/12] tests/vm: Add workaround to consume console
>   Date: Tue, 19 May 2020 09:22:59 -0400
>   Message-Id: <20200519132259.405-13-robert.foley@linaro.org>

Exactly! I guess Marc-André or Eduardo suggested the same fix (use a
dedicated polling thread).

> 
>>
>> Cleber has been working on some PoC / kludge.
>>
>> Cleber/Eduardo do you remember the problem?
>>
>>>
>>>>   (1/1) tests/acceptance/replay_kernel.py:ReplayKernel.test_x86_64_pc:
>>>> replay: recording...
>>>> replay: replaying...
>>>> INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout
>>>> reached\nOriginal status: ERROR\n{'name':
>>>> '1-tests/acceptance/replay_kernel.py:ReplayKernel.test_x86_64_pc',
>>>> 'logdir':
>>>> 'avocado/job-results/job-2020-05-27T16.48-71d7bf4/test-results/1-tes...
>>>> (90.68 s)
>>>>
>>>
> 
>