[kbuild-all] [radeon-alex:amd-staging-4.11 1061/1085] sparc64-linux-gnu-gcc: error: unrecognized command line option '-msse'; did you mean '-fdse'?

[Harry Wentland]

On 2017-05-16 05:56 AM, Christian König wrote:
> Am 16.05.2017 um 11:16 schrieb Michel Dänzer:
>> [ Dropping build robot aliases, adding amd-gfx and DC folks ]
>>
>> On 16/05/17 05:18 PM, Philip Li wrote:
>>> On Tue, May 16, 2017 at 10:04:32AM +0200, Christian König wrote:
>>>> Am 16.05.2017 um 07:20 schrieb zhoucm1:
>>>>> On 2017年05月15日 17:52, kbuild test robot wrote:
>>>>>> tree: git://people.freedesktop.org/~agd5f/linux.git amd-staging-4.11
>>>>>> head:   c285c73f2213f503a93aa142fff186e160b4a371
>>>>>> commit: 5f92704d0cc5928824ffa4ff8cec5cd4c0f55b24 [1061/1085]
>>>>>> drm/amd: fix init order of sched job
>>>>>> config: sparc-allmodconfig (attached as .config)
>>>>>> compiler: sparc64-linux-gnu-gcc (Debian 6.1.1-9) 6.1.1 20160705
>>>>>> reproduce:
>>>>>>          wget 
>>>>>> https://raw.githubusercontent.com/01org/lkp-tests/master/sbin/make.cross 
>>>>>>
>>>>>> -O ~/bin/make.cross
>>>>>>          chmod +x ~/bin/make.cross
>>>>>>          git checkout 5f92704d0cc5928824ffa4ff8cec5cd4c0f55b24
>>>>>>          # save the attached .config to linux build tree
>>>>>>          make.cross ARCH=sparc
>>>>>>
>>>>>> All errors (new ones prefixed by >>):
>>>>>>
>>>>>>>> sparc64-linux-gnu-gcc: error: unrecognized command line
>>>>>>>> option '-msse'; did you mean '-fdse'?
>>>>>>>> sparc64-linux-gnu-gcc: error: unrecognized command line
>>>>>>>> option '-mpreferred-stack-boundary=4'
>> [...]
>>
>>> Thanks for input, the issue does exist (as introduced by
>>> "drm/amdgpu/display: Enable DCN in DC"), so the report like
>>> https://lists.01.org/pipermail/kbuild-all/2017-May/033857.html
>>> does make sense.
>> It's because some DC Makefiles hardcode -msse and other compiler flags
>> which aren't supported on all architectures.
>>
>> DC folks, please fix this, either by only adding flags on architectures
>> where they're supported, or by preventing the corresponding code from
>> getting built on architectures where they're not supported.
> 
> Ah, yes that's indeed an issue.
> 
> But please at least fix the bisect or otherwise the wrong people get 
> those mails and wonder about the error messages.
> 

I just submitted a fix for this: "drm/amd/display: Limit DCN to x86 arch"

It seems to do the trick but I've never really dealt with multiple 
architectures before. Would appreciate if someone more knowledgeable 
than myself reviews this.

Essentially I'm blocking DCN code for non-x86 architectures for now.

Thanks,
Harry

> Christian.
>