[llvm-commits] [llvm] r66776 - in /llvm/trunk: lib/Target/X86/X86ISelLowering.cpp test/CodeGen/X86/i64-mem-copy.ll test/CodeGen/X86/mmx-copy-gprs.ll

[Evan Cheng]

It's safe. This transform is only done when load and stores are  
implemented with movsd.

Evan

On Mar 12, 2009, at 1:52 AM, Duncan Sands wrote:

> Hi Evan,
>
>> On x86, if the only use of a i64 load is a i64 store, generate a  
>> pair of double load and store instead.
>
> will this always work properly?  Maybe problems like those in PR2680  
> can occur?
>
> Ciao,
>
> Duncan.