peci-pcie CI issues

Andrew Geissler geissonator at gmail.com
Wed Jan 13 06:10:03 AEDT 2021



> On Jan 11, 2021, at 6:08 PM, Bills, Jason M <jason.m.bills at linux.intel.com> wrote:
> 
> 
> 
> On 12/24/2020 10:53 AM, Ed Tanous wrote:
>> On Thu, Dec 24, 2020 at 10:34 AM Andrei Kartashev <a.kartashev at yadro.com> wrote:
>>> 
>>> Well, then probably we can wait.
>>> How far this could happens?
>> Whenever the work gets done.  Someone needs to:
>> Send a patch to yocto upgrading the boost recipe.
>> Wait for the meta-layer bump to run (I think Andrew runs the job once a week).
>> Resolve any issues with the bump when it gets merged to OpenBMC.
>> There's no exact timelines on the above, but you can certainly
>> accelerate it by doing step 1, after which you're probably looking at
>> a couple weeks before we get it in OpenBMC.
> 
> It looks like upstream Yocto picked up boost 1.75: https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-support/boost.

Here’s our bump to try and pick it up: https://gerrit.openbmc-project.xyz/c/openbmc/openbmc/+/39533 

But we hit a build issue (unrelated to the boost aspect I believe).
https://github.com/openbmc/openbmc/issues/3746 

Help appreciated on the fail.
> 
>>> 
>>> On Thu, 2020-12-24 at 10:23 -0800, Ed Tanous wrote:
>>>> On Thu, Dec 24, 2020 at 10:07 AM Patrick Williams <patrick at stwcx.xyz>
>>>> wrote:
>>>>> We have had this issue with a number of repositories lately. The
>>>>> most recent version of boost::asio does not allow -fno-rtti.  The
>>>>> makefile needs to be changed to no longer force this option.
>>>> 
>>>> Or, as another option, just wait until boost 1.75.0 lands in yocto
>>>> master and subsequent openbmc bump.  It was released a couple weeks
>>>> ago and fixes this issue.  We'll likely be adding the no-rtti flags
>>>> back to most of the repos shortly after that.
>>>> 
>>>>> Sent from my iPhone
>>>>> 
>>>>>> On Dec 24, 2020, at 9:48 AM, Andrei Kartashev <
>>>>>> a.kartashev at yadro.com> wrote:
>>>>>> 
>>>>>> Hello Jason,
>>>>>> 
>>>>>> I push several patches to peci-pcie repo, but looks like CI
>>>>>> broken
>>>>>> there. Could you take a look on how to fix CI?
>>>>>> 
>>>>>> [ 90%] Building CXX object CMakeFiles/peci-
>>>>>> pcie.dir/src/peci_pcie.cpp.o
>>>>>> In file included from
>>>>>> /usr/local/include/boost/asio/execution.hpp:19,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/system_executor.hpp:20,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/associated_executor.hpp:22,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/detail/bind_handler.hpp:20,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/detail/wrapped_handler.hpp:18,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/io_context.hpp:23,
>>>>>>                 from
>>>>>> /usr/local/include/boost/asio/io_service.hpp:18,
>>>>>>                 from /home/jenkins-op/workspace/ci-
>>>>>> repository/openbmc/peci-pcie/src/peci_pcie.cpp:22:
>>>>>> /usr/local/include/boost/asio/execution/any_executor.hpp: In
>>>>>> static member function ‘static const std::type_info&
>>>>>> boost::asio::execution::detail::any_executor_base::target_type_vo
>>>>>> id()’:
>>>>>> /usr/local/include/boost/asio/execution/any_executor.hpp:811:23:
>>>>>> error: cannot use ‘typeid’ with ‘-fno-rtti’
>>>>>>  811 |     return typeid(void);
>>>>>>      |                       ^
>>>>>> /usr/local/include/boost/asio/execution/any_executor.hpp: In
>>>>>> static member function ‘static const std::type_info&
>>>>>> boost::asio::execution::detail::any_executor_base::target_type_ex
>>>>>> ()’:
>>>>>> /usr/local/include/boost/asio/execution/any_executor.hpp:851:21:
>>>>>> error: cannot use ‘typeid’ with ‘-fno-rtti’
>>>>>>  851 |     return typeid(Ex);
>>>>>>      |                     ^
>>>>>> 
>>>>>> --
>>>>>> Best regards,
>>>>>> Andrei Kartashev
>>>>>> 
>>>>>> 
>>> --
>>> Best regards,
>>> Andrei Kartashev
>>> 
>>> 



More information about the openbmc mailing list