AMDGPU: Fix scratch wave offset relative FI expansion

The offset may not be an inline immediate, so this needs
to be materialized into a register. The post-RA run of
SIShrinkInstructions is able to fold it later if it can.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@305761 91177308-0d34-0410-b5e6-96231b3b80d8
2 files changed