[PATCH] D28170: [LICM] When promoting scalars, allow inserting stores to thread-local allocas

[Xin Tong via llvm-commits]

I see.
Thanks
-Xin

On Wed, Jan 18, 2017 at 2:20 PM, Michael Kuperstein <mkuper at google.com> wrote:
> Assuming I understand you correctly - no, that's the mistake I made in the
> original version of this patch. :-)
> See discussion with Philip on https://reviews.llvm.org/D28147 .
>
> On Wed, Jan 18, 2017 at 2:16 PM, Xin Tong via Phabricator
> <reviews at reviews.llvm.org> wrote:
>>
>> trentxintong added inline comments.
>>
>>
>> ================
>> Comment at: llvm/trunk/lib/Transforms/Scalar/LICM.cpp:1037
>>      SafeToInsertStore =
>> -        isAllocLikeFn(Object, TLI) && !PointerMayBeCaptured(Object, true,
>> true);
>> +        (isAllocLikeFn(Object, TLI) || isa<AllocaInst>(Object)) &&
>> +        !PointerMayBeCaptured(Object, true, true);
>> ----------------
>> The question I really want to ask is whether the check for
>> MayNotBeCaptured is necessary for Alloca. Should not alloca be implicitly
>> thread local ? The language reference does not explicitly state it.
>>
>>
>> Repository:
>>   rL LLVM
>>
>> https://reviews.llvm.org/D28170
>>
>>
>>
>