forked from OSchip/llvm-project
parent
afff169f17
commit
a2010755fa
|
@ -183,8 +183,8 @@ so cool to turn it into something like:
|
|||
|
||||
... which would only do one 32-bit XOR per loop iteration instead of two.
|
||||
|
||||
It would also be nice to recognize the reg->size doesn't alias reg->node[i], but
|
||||
this requires TBAA.
|
||||
It would also be nice to recognize the reg->size doesn't alias reg->node[i],
|
||||
but this requires TBAA.
|
||||
|
||||
//===---------------------------------------------------------------------===//
|
||||
|
||||
|
|
Loading…
Reference in New Issue