[all-commits] [llvm/llvm-project] 17aa37: [SelectionDAG] Add memory size for CSEMap ID calcu...

[Alex MacLean via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 17aa37dd30ca2d0b85a151c0c8102946a2cdee4b
      https://github.com/llvm/llvm-project/commit/17aa37dd30ca2d0b85a151c0c8102946a2cdee4b
  Author: Alex MacLean <amaclean at nvidia.com>
  Date:   2023-06-26 (Mon, 26 Jun 2023)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
    A llvm/test/CodeGen/NVPTX/dag-cse.ll

  Log Message:
  -----------
  [SelectionDAG] Add memory size for CSEMap ID calculation

In NVPTX `ReplaceVectorLoad()`, i1 and i8 types are promoted to i16,
followed by a truncate operation. Thus, v2i8 (or v2i1) and v2i16 will
have the same VTList, which causes a collision in CSEMap.

To differentiate the original VTList, let's add the size in generating
an ID. Otherwise the compiler crashes in refineAlignment:
`MMO->getSize() == getSize() && "Size mismatch!"`

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D153712