On 11/27/19 6:07 PM, Jan Kiszka wrote:
> On 27.11.19 17:48, Philippe Gerum wrote:
>> On 11/26/19 6:38 PM, Jan Kiszka via Xenomai wrote:
>>> On 26.11.19 18:29, Jeff Webb wrote:
>>>> On 11/15/19 5:14 PM, Jeff Webb via Xenomai wrote:
>>>>> On Friday, November 15, 2019 4:40 AM, Jan Kiszka
>>>>> <[email protected]> wrote:
>>>>>> On 15.11.19 01:31, Jeff Webb via Xenomai wrote:
>>>>>>> I am building a Xenomai cobalt kernel under Ubuntu 18.04 based on:
>>>>>>>
>>>>>>>        Linux 4.19.75
>>>>>>>        Xenomai 3.1-rc3
>>>>>>>        ipipe-core-4.19.75-x86-7.patch
>>>>>>>
>>>>>>>
>>>>>>> After running prepare-kernel.sh and "makeolddefconfig" (using the
>>>>>>> unmodified default .config), I get this error:
>>>>>>>
>>>>>>>        WARNING: unmet direct dependencies detected for PARAVIRT
>>>>>>>          Depends on [n]: HYPERVISOR_GUEST [=y] && !IPIPE [=y]
>>>>>>>          Selected by [m]:
>>>>>>>          - HYPERV [=m] && X86 [=y] && ACPI [=y] && X86_LOCAL_APIC [=y]
>>>>>>> && HYPERVISOR_GUEST [=y]
>>>>>>>
>>>>>>>
>>>>>>> I turned off HYPERVISOR_GUEST manually to fix the warning. I hadn't
>>>>>>> seen this type of error before, but perhaps something could be done in
>>>>>>> the patch to make the default config consistent... Just FYI.
>>>>>>
>>>>>> It's HYPERV what causes this, by also pulling PARAVIRT which is not
>>>>>> compatible with I-pipe. We need to tag that config as well to avoid this.
>>>>>>
>>>>>>> Otherwise, everything went fine with the build process and initial
>>>>>>> testing. I will be giving this build some more testing over the next
>>>>>>> week or so and will let you know if I find any issues.
>>>>>>
>>>>>> Great to hear, thanks.
>>>>>
>>>>> I haven't done extensive testing yet, but I tried several of our
>>>>> real-time applications with the new build, and things seem to be working
>>>>> so far.  Even on a different (but perhaps similar), older set of
>>>>> hardware from the one that has the intremap issues, I get the same
>>>>> kernel BUG messages (using the same 4.19 kernel build) mentioned in my
>>>>> other email:
>>>>>
>>>>>      https://xenomai.org/pipermail/xenomai/2019-November/041988.html
>>>>>
>>>>> but things appear to be functioning properly as far as I can tell.
>>>>>
>>>>> This is just some more feedback on the latest kernel / xenomai release
>>>>> candidate.
>>>>
>>>> All is going well so far with Xenomai v3.1-rc3 /
>>>> ipipe-core-4.19.84-x86-8.patch.  Given the recent patches posted to the
>>>> list, are there plans for another RC, or will the next release be the
>>>> official 3.1?
>>>>
>>>
>>> Thanks for validating!
>>>
>>> The scheduler changes qualify for another -rc, yes, I'm just waiting for an
>>> update from Philippe for a closing patch on this topic.
>>>
>>
>> Patiently reviewing all call sites kicking the rescheduling procedure here.
>> The picture does not look as simple as I originally thought. No blocker,
>> but a bit more work to prevent the root stage from rescheduling with hard
>> irqs on.
>>
> 
> Well, I have my own blocker now anyway (__SCHED_CURRENT removal)...
> 

With the v2 series, I'm done with the reschedule-over-root mess. Non-regression
tests are happy, switchtest too. This definitely calls for another -rc though.

-- 
Philippe.

Reply via email to