[llvm-commits] [PATCH] Try to figure out <size> argument in llvm.lifetime intrinsics

[Alexey Samsonov]

r167821, thanks!

On Tue, Nov 13, 2012 at 3:34 AM, Nick Lewycky <nlewycky at google.com> wrote:

> 675        if (IFI.TD != 0) {
>
> The rest of the file and llvm use "if (IFI.TD)" style. Please fix.
>
> LGTM!
>
> On 12 November 2012 01:22, Alexey Samsonov <samsonov at google.com> wrote:
>
>>
>>   Kostya is right about merging allocas - stack reuse won't be able to do
>> anything if ASan is enabled.
>>
>>
>> ================
>> Comment at: lib/Transforms/Utils/InlineFunction.cpp:676
>> @@ +675,3 @@
>> +        if (IFI.TD != 0) {
>> +          uint64_t SizeValue = AllocaArraySize->getLimitedValue();
>> +          Type *AllocaType = AI->getAllocatedType();
>> ----------------
>> Nick Lewycky wrote:
>> > What if the array size is an i128, and greater than fits in a uint64_t?
>> getLimitedValue() will return -1 (~0ULL). Maybe you want to bail in that
>> case?
>> Done
>>
>> ================
>> Comment at: lib/Transforms/Utils/InlineFunction.cpp:678
>> @@ +677,3 @@
>> +          Type *AllocaType = AI->getAllocatedType();
>> +          SizeValue *= IFI.TD->getTypeAllocSize(AllocaType);
>> +          AllocaSize =
>> ConstantInt::get(Type::getInt64Ty(AI->getContext()),
>> ----------------
>> Nick Lewycky wrote:
>> > What about overflow here?
>> >
>> > I don't know of any shared utility to multiply and detect overflow, but
>> I've needed to do the exact same thing before. See umul_ov() at
>> ScalarEvolution.cpp:1826.
>> >
>> Done
>>
>>
>> http://llvm-reviews.chandlerc.com/D110
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>
>


-- 
Alexey Samsonov, MSK
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20121113/f97893bc/attachment.html>