[LLVMdev] RFC: Missing canonicalization in LLVM

[Krzysztof Parzyszek]

On 1/21/2015 4:16 PM, Chandler Carruth wrote:
>
> 2) It doesn't effectively move us toward the optimizer treating IR with
> different pointee types for pointer types indistinguishably. Some day, I
> continue to think we should get rid of the pointee types entirely.

Pointee?  I'd think that eliminating multiple pointer types would be a 
better thing to "normalize". Having a tag on a load stating the loaded 
type would be useful for instruction selection.  Unless I'm not reading 
this correctly...

-Krzysztof


-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, 
hosted by The Linux Foundation