[Patch] Enable value forwarding for loads from calloc without intervening store

Philip Reames listmail at philipreames.com
Thu Jun 12 09:11:25 PDT 2014


On 06/12/2014 08:16 AM, Duncan P. N. Exon Smith wrote:
>> On 2014 Jun 11, at 10:15, Philip Reames <listmail at philipreames.com> wrote:
>>
>> Attached.
> Thanks.
>
>> http://reviews.llvm.org/D3931
>>
>> Philip
>> <calloc-load-forward.diff>
> LGTM.  Do you need me to commit it for you?
I don't have commit access, so please do.
>
> BTW, I noticed another strange whitespace thing -- if I'm committing it,
> I'll just fix this myself, but if you're committing it please run
> clang-format:
>
> +      ValuesPerBlock.push_back(AvailableValueInBlock::get(DepBB,
> +                                             Constant::getNullValue(LI->getType())));
>
> This indentation is strange (and violates 80-columns).  clang-format's
> opinion is this:
>
>        ValuesPerBlock.push_back(AvailableValueInBlock::get(
>            DepBB, Constant::getNullValue(LI->getType())));
Will do for future commits.  I don't remember exactly why I picked that 
format; it does look odd.  I think I may have coped that formatting from 
the malloc case above it.  I was probably trying to keep consistency.

Philip




More information about the llvm-commits mailing list