[PATCH] D52683: [AMDGPU] Fix for negative offsets in buffer/tbuffer intrinsics

[Tim Renouf via Phabricator via llvm-commits]

tpr added a comment.

For the old intrinsics, there used to be code called from dag pattern matching to convert a single offset into some combination of voffset, soffset and instoffset. Is that what you're referring to?

I replaced that with this code here, earlier on in ISelLowering when converting the intrinsic to a DAG node, that converts an offset into some combination of voffset and instoffset.


Repository:
  rL LLVM

https://reviews.llvm.org/D52683