a56993a694
It looks like `+promote-alloca` is always enabled, and `-promote-alloca` is simply used as a switch to toggle the pass.
17 lines
617 B
LLVM
17 lines
617 B
LLVM
; RUN: llc -mtriple=amdgcn -mcpu=tahiti -disable-promote-alloca-to-vector -disable-promote-alloca-to-lds < %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=tonga -disable-promote-alloca-to-vector -disable-promote-alloca-to-lds < %s
|
|
|
|
; Test that INSERT_SUBREG instructions don't have non-register operands after
|
|
; instruction selection.
|
|
|
|
; Make sure this doesn't crash
|
|
; CHECK-LABEL: test:
|
|
define amdgpu_kernel void @test(ptr addrspace(1) %out) {
|
|
entry:
|
|
%tmp0 = alloca [16 x i32], addrspace(5)
|
|
%tmp1 = ptrtoint ptr addrspace(5) %tmp0 to i32
|
|
%tmp2 = sext i32 %tmp1 to i64
|
|
store i64 %tmp2, ptr addrspace(1) %out
|
|
ret void
|
|
}
|