+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck %s
; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca -disable-promote-alloca-to-lds< %s | FileCheck -check-prefix=NOLDS %s
; This normally would be fixed by instcombine to be compare to the GEP
; indices
-; NOLDS-NOT: addrspace(3)
-
-; CHECK-LABEL: @lds_promoted_alloca_icmp_same_derived_pointer(
-; CHECK: [[ARRAYGEP:%[0-9]+]] = getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promoted_alloca_icmp_same_derived_pointer.alloca, i32 0, i32 %{{[0-9]+}}
-; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %a
-; CHECK: %ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %b
-; CHECK: %cmp = icmp eq i32 addrspace(3)* %ptr0, %ptr1
define amdgpu_kernel void @lds_promoted_alloca_icmp_same_derived_pointer(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
+; CHECK-LABEL: @lds_promoted_alloca_icmp_same_derived_pointer(
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; CHECK-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; CHECK-NEXT: [[PTR1:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[B:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], [[PTR1]]
+; CHECK-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; CHECK-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; CHECK-NEXT: ret void
+;
+; NOLDS-LABEL: @lds_promoted_alloca_icmp_same_derived_pointer(
+; NOLDS-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; NOLDS-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; NOLDS-NEXT: [[PTR1:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[B:%.*]]
+; NOLDS-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], [[PTR1]]
+; NOLDS-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; NOLDS-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; NOLDS-NEXT: ret void
+;
%alloca = alloca [16 x i32], align 4
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %a
%ptr1 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %b
ret void
}
-; CHECK-LABEL: @lds_promoted_alloca_icmp_null_rhs(
-; CHECK: [[ARRAYGEP:%[0-9]+]] = getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promoted_alloca_icmp_null_rhs.alloca, i32 0, i32 %{{[0-9]+}}
-; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %a
-; CHECK: %cmp = icmp eq i32 addrspace(3)* %ptr0, null
define amdgpu_kernel void @lds_promoted_alloca_icmp_null_rhs(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
+; CHECK-LABEL: @lds_promoted_alloca_icmp_null_rhs(
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; CHECK-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], null
+; CHECK-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; CHECK-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; CHECK-NEXT: ret void
+;
+; NOLDS-LABEL: @lds_promoted_alloca_icmp_null_rhs(
+; NOLDS-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; NOLDS-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; NOLDS-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], null
+; NOLDS-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; NOLDS-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; NOLDS-NEXT: ret void
+;
%alloca = alloca [16 x i32], align 4
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %a
%cmp = icmp eq i32* %ptr0, null
ret void
}
-; CHECK-LABEL: @lds_promoted_alloca_icmp_null_lhs(
-; CHECK: [[ARRAYGEP:%[0-9]+]] = getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promoted_alloca_icmp_null_lhs.alloca, i32 0, i32 %{{[0-9]+}}
-; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %a
-; CHECK: %cmp = icmp eq i32 addrspace(3)* null, %ptr0
define amdgpu_kernel void @lds_promoted_alloca_icmp_null_lhs(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
+; CHECK-LABEL: @lds_promoted_alloca_icmp_null_lhs(
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; CHECK-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32* null, [[PTR0]]
+; CHECK-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; CHECK-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; CHECK-NEXT: ret void
+;
+; NOLDS-LABEL: @lds_promoted_alloca_icmp_null_lhs(
+; NOLDS-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; NOLDS-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; NOLDS-NEXT: [[CMP:%.*]] = icmp eq i32* null, [[PTR0]]
+; NOLDS-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; NOLDS-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; NOLDS-NEXT: ret void
+;
%alloca = alloca [16 x i32], align 4
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %a
%cmp = icmp eq i32* null, %ptr0
ret void
}
-; CHECK-LABEL: @lds_promoted_alloca_icmp_unknown_ptr(
-; CHECK: %alloca = alloca [16 x i32], align 4
-; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %a
-; CHECK: %ptr1 = call i32* @get_unknown_pointer()
-; CHECK: %cmp = icmp eq i32* %ptr0, %ptr1
define amdgpu_kernel void @lds_promoted_alloca_icmp_unknown_ptr(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
+; CHECK-LABEL: @lds_promoted_alloca_icmp_unknown_ptr(
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; CHECK-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; CHECK-NEXT: [[PTR1:%.*]] = call i32* @get_unknown_pointer()
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], [[PTR1]]
+; CHECK-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; CHECK-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; CHECK-NEXT: ret void
+;
+; NOLDS-LABEL: @lds_promoted_alloca_icmp_unknown_ptr(
+; NOLDS-NEXT: [[ALLOCA:%.*]] = alloca [16 x i32], align 4
+; NOLDS-NEXT: [[PTR0:%.*]] = getelementptr inbounds [16 x i32], [16 x i32]* [[ALLOCA]], i32 0, i32 [[A:%.*]]
+; NOLDS-NEXT: [[PTR1:%.*]] = call i32* @get_unknown_pointer()
+; NOLDS-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR0]], [[PTR1]]
+; NOLDS-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+; NOLDS-NEXT: store volatile i32 [[ZEXT]], i32 addrspace(1)* [[OUT:%.*]], align 4
+; NOLDS-NEXT: ret void
+;
%alloca = alloca [16 x i32], align 4
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 %a
%ptr1 = call i32* @get_unknown_pointer()