Re: [PATCH] tests/vm/freebsd: Increase code coverage

[Philippe Mathieu-Daudé]

On 6/15/21 19:40, Wainer dos Santos Moschetta wrote:
> Hi,
> 
> On 6/15/21 2:19 PM, Daniel P. Berrangé wrote:
>> On Tue, Jun 15, 2021 at 06:02:02PM +0100, Daniel P. Berrangé wrote:
>>> On Mon, May 31, 2021 at 05:53:25PM -0300, Wainer dos Santos Moschetta
>>> wrote:
>>>> Hi,
>>>>
>>>> On 5/31/21 7:03 AM, Philippe Mathieu-Daudé wrote:
>>>>> Install more dependencies to increase code coverage.
>>>>>
>>>>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>>>>> ---
>>>>>    tests/vm/freebsd | 5 +++++
>>>>>    1 file changed, 5 insertions(+)
>>>> With or without this patch I got an error when `make
>>>> vm-build-freebsd`. It
>>>> fails to install packages.
>>>>
>>>> For example, with this patch I got:
>>>>
>>>> < Output omitted>
>>>>
>>>> ### Installing packages ...
>>>> Failed to prepare guest environment
>>>> Traceback (most recent call last):
>>>>    File "/home/wmoschet/src/qemu/tests/vm/basevm.py", line 634, in main
>>>>      return vm.build_image(args.image)
>>>>    File "/home/wmoschet/src/qemu/tests/vm/freebsd", line 206, in
>>>> build_image
>>>>      self.ssh_root_check("pkg install -y %s\n" % " ".join(self.pkgs))
>>>>    File "/home/wmoschet/src/qemu/tests/vm/basevm.py", line 255, in
>>>> ssh_root_check
>>>>      self._ssh_do(self._config["root_user"], cmd, True)
>>>>    File "/home/wmoschet/src/qemu/tests/vm/basevm.py", line 242, in
>>>> _ssh_do
>>>>      raise Exception("SSH command failed: %s" % cmd)
>>>> Exception: SSH command failed: pkg install -y git pkgconf bzip2
>>>> python37
>>>> ninja bash gmake gsed gettext cyrus-sasl gnutls nettle jpeg-turbo
>>>> png sdl2
>>>> gtk3 libxkbcommon pixman libepoxy mesa-libs zstd usbredir
>>>>
>>>> Is it a known issue?
>>> Hard to actually tell what the error really is. This message is
>>> only giving the command that was invoked, but seems to have thrown
>>> away stdout/stderr which would have the messages telling us what
>>> went wrong.  This lack of error reporting in basevm.py so badly
>>> needs to be fixed, otherwise we're working blind when debugging
>>> failures.
>> Hmm, when I try the same command, it *does* print all the output
>> so you can see what's going on, but it doesn't actually fail for
>> me either (without this patch).
> 
> Thomas sent me a message in private while ago that the error could be
> related with the amount of ssh keys in my agent-ssh. I didn' t check
> that hypothesis yet; will debug a littler further and let you all posted.
> 
> ah, if this problem cannot be reproduced on other machines....I see no
> reason to hold this patch.

OK thank you!