[LLVMdev] proposed new rule for getelementptr

[Dan Gohman]

On Jul 22, 2009, at 2:17 PM, Török Edwin wrote:


> On 2009-07-23 00:02, Dan Gohman wrote:
>
>> -fno-strict-aliasing is not impacted.
>>
>>
>>
>
> Will LLVM's default behavior be that of -fstrict-aliasing, or
> -fno-strict-aliasing?
> We don't have TBAA, but the "computer pointer doesn't alias existing
> pointers" sounds like something for -fno-strict-aliasing, no?
> I'm not sure what gcc does here, does -fno-strict-aliasing have any
> effect on pointer computed from ints?

It does sound like it from the name, but no, it's unrelated.

Dan