[llvm] r234466 - Not all triples put _ before function names. Specify a triple to make this test pass on Linux.

[Lang Hames]

Haven't had time to think about this carefully yet, but seems like a nice
clean option. I might run a sed over my regression set directory and see
what explodes. :)

- Lang.

On Thu, Apr 9, 2015 at 12:26 PM, Eric Christopher <echristo at gmail.com>
wrote:

>
>
> On Thu, Apr 9, 2015 at 10:45 AM Eric Christopher <echristo at gmail.com>
> wrote:
>
>> On Wed, Apr 8, 2015 at 11:02 PM Lang Hames <lhames at gmail.com> wrote:
>>
>>> Random underscores keep people on their toes. Except when they don't and
>>> you break all the builders and Nick has to bail you out. But I mean, at
>>> least Nick is on his toes.
>>>
>>> Slightly more seriously, triples seem to be the usual way to pin down
>>> regression test output. In general, compiling for the host might change the
>>> output in ways that aren't easily captured by patterns, even though they'd
>>> work here.
>>>
>>>
>> Yeah, I completely disagree with this entire statement. If something is
>> particularly host dependent or has elements of host dependence that could
>> make a test brittle, etc. In general, code gen tests shouldn't require a
>> triple to make it work.
>>
>>
> A compromise position btw here would be a triple of <arch>-unknown-unknown
> which would stabilize things (assembler, mangling, etc), but wouldn't have
> any particular OS initialization support etc.
>
> -eric
>
>
>> -eric
>>
>>
>>
>>> - Lang.
>>>
>>> On Wed, Apr 8, 2015 at 10:50 PM, Eric Christopher <echristo at gmail.com>
>>> wrote:
>>>
>>>> Relax the check maybe?
>>>>
>>>> On Wed, Apr 8, 2015, 10:40 PM Lang Hames <lhames at gmail.com> wrote:
>>>>
>>>>> Beat me to it. Thanks for the fix. :)
>>>>>
>>>>> - Lang.
>>>>>
>>>>> On Wed, Apr 8, 2015 at 10:31 PM, Nick Lewycky <nicholas at mxc.ca> wrote:
>>>>>
>>>>>> Author: nicholas
>>>>>> Date: Thu Apr  9 00:31:32 2015
>>>>>> New Revision: 234466
>>>>>>
>>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=234466&view=rev
>>>>>> Log:
>>>>>> Not all triples put _ before function names. Specify a triple to make
>>>>>> this test pass on Linux.
>>>>>>
>>>>>> Modified:
>>>>>>     llvm/trunk/test/CodeGen/AArch64/arm64-misaligned-memcpy-inline.ll
>>>>>>
>>>>>> Modified: llvm/trunk/test/CodeGen/AArch64/arm64-misaligned-
>>>>>> memcpy-inline.ll
>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/
>>>>>> CodeGen/AArch64/arm64-misaligned-memcpy-inline.ll?
>>>>>> rev=234466&r1=234465&r2=234466&view=diff
>>>>>> ============================================================
>>>>>> ==================
>>>>>> --- llvm/trunk/test/CodeGen/AArch64/arm64-misaligned-memcpy-inline.ll
>>>>>> (original)
>>>>>> +++ llvm/trunk/test/CodeGen/AArch64/arm64-misaligned-memcpy-inline.ll
>>>>>> Thu Apr  9 00:31:32 2015
>>>>>> @@ -1,4 +1,4 @@
>>>>>> -; RUN: llc -march=arm64 -aarch64-strict-align < %s | FileCheck %s
>>>>>> +; RUN: llc -march=arm64 -mtriple=arm64-apple-ios
>>>>>> -aarch64-strict-align < %s | FileCheck %s
>>>>>>
>>>>>>  ; Small (16-bytes here) unaligned memcpys should stay memcpy calls if
>>>>>>  ; strict-alignment is turned on.
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> llvm-commits mailing list
>>>>>> llvm-commits at cs.uiuc.edu
>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> llvm-commits mailing list
>>>>> llvm-commits at cs.uiuc.edu
>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>>>
>>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150409/48f51bd5/attachment.html>