kicad-developers team mailing list archive

["Steven A. Falco" <stevenfalco@xxxxxxxxx>]

On 8/27/19 11:14 AM, Tomasz Wlostowski wrote:
> On 27/08/2019 16:18, Steven A. Falco wrote:
>> On 8/26/19 4:16 PM, Seth Hillbrand wrote:
>>> On 2019-08-26 16:05, Tomasz Wlostowski wrote:
>>>> On 26/08/2019 17:24, Seth Hillbrand wrote:
>>>>> Agreed.  That looks like it may be have been a rebase issue as the
>>>>> commit was for MSVC and shouldn't affect PPC.
>>>>>
>>>>> @Tom, shout if you see an issue here.  I've pushed the patch to master
>>>>> in the meantime.
>>>>>
>>>> I don't see a technical issue, but rather a practical one: why Linux
>>>> distributions dictate us what architectures we should support?
>>>>
>>>> I've never seen a machine with a little-endian PPC.
>>>>
>>>> Tom
>>>
>>> The LE variant is the PPC future starting with POWER8[1][2].
>>>
>>> [1] https://www.phoronix.com/scan.php?page=news_item&px=POWER-PPC64-Discontinue-Fedora
>>> [2] https://developer.ibm.com/articles/l-power-little-endian-faq-trs/
>>>
>>
>> I don't think it is a question of LE vs BE.  The issue is 32-bit vs 64-bit.
>>
>> On a 64-bit PPC machine, gcc defines both _ARCH_PPC and _ARCH_PPC64, so the test in libcontext.h was wrong, and resulted in both 32-bit and 64-bit code trying to be compiled simultaneously.  My patch simply makes it "either 32bit or 64-bit" rather than "both 32-bit and 64-bit at the same time".
>>
> 
> Does anybody here have a PPC machine to test this? Preferably one with
> working OpenGL?
> 
> Cheers,
> Tom
> 

I actually tested it on a PPC64LE virtual machine.  While it is naturally slow on a VM, I was able to launch KiCAD with my patch, and it worked perfectly.

	Steve