Re: [Xen-devel] [xen-unstable test] 77945: regressions - FAIL

[Doug Goldstein <cardoe@xxxxxxxxxx>]

On 1/14/16 8:44 AM, Ian Campbell wrote:
> On Thu, 2016-01-14 at 08:07 -0600, Doug Goldstein wrote:
>> On 1/14/16 7:57 AM, Ian Campbell wrote:
>>> On Thu, 2016-01-14 at 05:50 -0700, Jan Beulich wrote:
>>>>>>> On 14.01.16 at 12:42, <ian.campbell@xxxxxxxxxx> wrote:
>>>>> The bisector is working:
>>>>> http://logs.test-lab.xenproject.org/osstest/results/bisect/xen-unst
>>>>> able
>>>>> /test 
>>>>> -amd64-i386-xl-qemut-stubdom-debianhvm-amd64-xsm.xen-install.html
>>>>>
>>>>> and has gotten to this range:
>>>>>
>>>>>     $ git log --oneline 5513bd0b4675..20c8f1a8a5fd
>>>>>     20c8f1a convert XSM_ENABLE to Kconfig
>>>>>     529298f convert FLASK_ENABLE to Kconfig
>>>>>     361b4f9 build: save generated xen .config
>>>>>     e3cce17 x86/VPMU: implement ipc and arch filter flags
>>>>>
>>>>> I suspect the issue is that 529298f removes FLASK_ENABLE from the
>>>>> top-
>>>>> level, but that is used in tools/Makefile. i.e. there is that other
>>>>> patch
>>>>> which needed to go before the two included above.
>>>>
>>>> And is that other patch ready to go in? If not, do we need to
>>>> revert another time?
>>>
>>> I think that is likely to be the case, yes.
>>>
>>>> In any event - Doug, you should point out such dependencies in
>>>> the submission, e.g. after the first --- marker.
>>>
>>> Really they should have been in the same series in this case.
>>>
>>> Ian.
>>>
>>
>> They were all in the same thread. And were all acked. But the first one
>> against the tools shouldn't have been necessary, it should have still
>> built.
> 
> I was talking about "tools: make flask utils build unconditional" from <145
> 0759603-24249-1-git-send-email-cardoe@xxxxxxxxxx> which appears to be a
> singleton patch, not part of a series. Maybe that is unrelated to this
> breakage then?
> 
> Ian.
> 
> 

It shouldn't have been required. Its all controlled by
--enable-xsmpolicy or --disable-xsmpolicy.

But I now see where it is. Technically the OSSTest depends on
FLASK_ENABLE but only sets XSM_ENABLE in the tests. It relies on the
fact that FLASK_ENABLE defaults to the same value as XSM_ENABLE but I
removed that in the last patch.

I will provide patches to OSSTest to set the proper variables rather
than assuming they will be populated appropriately.

-- 
Doug Goldstein

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel