[libc++ patch] Make meta.trans.other/aligned_storage.pass.cpp pass on arm

[Nico Weber]

Ping.


On Fri, May 30, 2014 at 4:43 PM, Nico Weber <thakis at chromium.org> wrote:

> The attached patch implements your suggestion (without the max). I
> verified that this passes on x86 os x and arm android.
>
>
> On Wed, May 28, 2014 at 4:25 PM, Nico Weber <thakis at chromium.org> wrote:
>
>> On Wed, May 28, 2014 at 4:04 PM, Marshall Clow <mclow.lists at gmail.com>
>> wrote:
>>
>>>
>>> On May 28, 2014, at 4:36 AM, Nico Weber <thakis at chromium.org> wrote:
>>>
>>> > Hi,
>>> >
>>> > On arm, the maxium alignment is 8. The attached patch tweaks
>>> meta.trans.other/aligned_storage.pass.cpp so that it passes on arm. (The
>>> test currently assumes that alignment goes up to at least 16.)
>>>
>>> Nico —
>>>
>>> I’m a bit leery of
>>>         + static_assert(std::alignment_of<T1>::value == alignof(T1), "");
>>> b/c I’m not sure that it tests what we want to test here.
>>>
>>> Is there some way that we can use max_align_t in this test?
>>>
>>> Maybe something like (untested code):
>>>         static_assert (  std::alignment_of<T1>::value == std::max(16,
>>> alignof(std::max_align_t));
>>>
>>
>> Why the max? On arm, alignof(max_align_t) is 8 (just like alignof(T1)),
>> so max(16, alignof(max_aling_t)) is 16, while std::alignment_of<T1>::value
>> is 8.
>>
>>     static_assert(std::alignment_of<T1>::value ==
>>                   alignof(std::max_align_t), "");
>>
>> does work though, if you like that better. Should I just
>> s/alignof(T1)/alignof(std::max_align_t)/ in my patch?
>>
>>
>>>
>>> — Marshall
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140602/946a2706/attachment.html>