AMDGPU: Fix assert on m0_lo16/m0_hi16
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Mon, 14 Jun 2021 16:53:36 +0000 (12:53 -0400)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Fri, 18 Jun 2021 22:48:53 +0000 (18:48 -0400)
commitad4a18251a37a244e3d4375538abfc0894a1e6ac
tree03c5112c50e71009f7ba6feb9ddb8bbf3deafca1
parentec978664542d6202c05e809fcfd6ead62a43d41b
AMDGPU: Fix assert on m0_lo16/m0_hi16

These get added (redundantly) to the bundle expanded for indirect
register accesses. We hit this path only when there is a call in the
function.
llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
llvm/test/CodeGen/AMDGPU/indirect-addressing-si-gfx9.ll