diff options
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll')
-rw-r--r-- | llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll | 104 |
1 files changed, 52 insertions, 52 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll b/llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll index ee47b14c496..8cf1e16c96d 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.u32.ll @@ -8,7 +8,7 @@ declare i32 @llvm.AMDGPU.bfe.u32(i32, i32, i32) nounwind readnone ; FUNC-LABEL: {{^}}bfe_u32_arg_arg_arg: ; SI: v_bfe_u32 ; EG: BFE_UINT -define void @bfe_u32_arg_arg_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src1, i32 %src2) nounwind { +define amdgpu_kernel void @bfe_u32_arg_arg_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src1, i32 %src2) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 %src0, i32 %src1, i32 %src1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -17,7 +17,7 @@ define void @bfe_u32_arg_arg_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src1, i ; FUNC-LABEL: {{^}}bfe_u32_arg_arg_imm: ; SI: v_bfe_u32 ; EG: BFE_UINT -define void @bfe_u32_arg_arg_imm(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { +define amdgpu_kernel void @bfe_u32_arg_arg_imm(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 %src0, i32 %src1, i32 123) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -26,7 +26,7 @@ define void @bfe_u32_arg_arg_imm(i32 addrspace(1)* %out, i32 %src0, i32 %src1) n ; FUNC-LABEL: {{^}}bfe_u32_arg_imm_arg: ; SI: v_bfe_u32 ; EG: BFE_UINT -define void @bfe_u32_arg_imm_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src2) nounwind { +define amdgpu_kernel void @bfe_u32_arg_imm_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src2) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 %src0, i32 123, i32 %src2) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -35,7 +35,7 @@ define void @bfe_u32_arg_imm_arg(i32 addrspace(1)* %out, i32 %src0, i32 %src2) n ; FUNC-LABEL: {{^}}bfe_u32_imm_arg_arg: ; SI: v_bfe_u32 ; EG: BFE_UINT -define void @bfe_u32_imm_arg_arg(i32 addrspace(1)* %out, i32 %src1, i32 %src2) nounwind { +define amdgpu_kernel void @bfe_u32_imm_arg_arg(i32 addrspace(1)* %out, i32 %src1, i32 %src2) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 123, i32 %src1, i32 %src2) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -45,7 +45,7 @@ define void @bfe_u32_imm_arg_arg(i32 addrspace(1)* %out, i32 %src1, i32 %src2) n ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_arg_0_width_reg_offset(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { +define amdgpu_kernel void @bfe_u32_arg_0_width_reg_offset(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 %src0, i32 %src1, i32 0) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -55,7 +55,7 @@ define void @bfe_u32_arg_0_width_reg_offset(i32 addrspace(1)* %out, i32 %src0, i ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_arg_0_width_imm_offset(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { +define amdgpu_kernel void @bfe_u32_arg_0_width_imm_offset(i32 addrspace(1)* %out, i32 %src0, i32 %src1) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 %src0, i32 8, i32 0) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -65,7 +65,7 @@ define void @bfe_u32_arg_0_width_imm_offset(i32 addrspace(1)* %out, i32 %src0, i ; SI: buffer_load_ubyte ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_zextload_i8(i32 addrspace(1)* %out, i8 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zextload_i8(i32 addrspace(1)* %out, i8 addrspace(1)* %in) nounwind { %load = load i8, i8 addrspace(1)* %in %ext = zext i8 %load to i32 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %ext, i32 0, i32 8) @@ -82,7 +82,7 @@ define void @bfe_u32_zextload_i8(i32 addrspace(1)* %out, i8 addrspace(1)* %in) n ; VI-NEXT: v_and_b32_e32 ; SI-NOT: {{[^@]}}bfe ; GCN: s_endpgm -define void @bfe_u32_zext_in_reg_i8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 255 @@ -97,7 +97,7 @@ define void @bfe_u32_zext_in_reg_i8(i32 addrspace(1)* %out, i32 addrspace(1)* %i ; SI-NEXT: v_and_b32_e32 ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_zext_in_reg_i16(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i16(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 65535 @@ -111,7 +111,7 @@ define void @bfe_u32_zext_in_reg_i16(i32 addrspace(1)* %out, i32 addrspace(1)* % ; SI: v_add_i32 ; SI: bfe ; SI: s_endpgm -define void @bfe_u32_zext_in_reg_i8_offset_1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i8_offset_1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 255 @@ -126,7 +126,7 @@ define void @bfe_u32_zext_in_reg_i8_offset_1(i32 addrspace(1)* %out, i32 addrspa ; SI-NEXT: v_and_b32_e32 {{v[0-9]+}}, 0xf8 ; SI-NEXT: bfe ; SI: s_endpgm -define void @bfe_u32_zext_in_reg_i8_offset_3(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i8_offset_3(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 255 @@ -141,7 +141,7 @@ define void @bfe_u32_zext_in_reg_i8_offset_3(i32 addrspace(1)* %out, i32 addrspa ; SI-NEXT: v_and_b32_e32 {{v[0-9]+}}, 0x80 ; SI-NEXT: bfe ; SI: s_endpgm -define void @bfe_u32_zext_in_reg_i8_offset_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i8_offset_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 255 @@ -155,7 +155,7 @@ define void @bfe_u32_zext_in_reg_i8_offset_7(i32 addrspace(1)* %out, i32 addrspa ; SI: v_add_i32 ; SI-NEXT: bfe ; SI: s_endpgm -define void @bfe_u32_zext_in_reg_i16_offset_8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_zext_in_reg_i16_offset_8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %load = load i32, i32 addrspace(1)* %in, align 4 %add = add i32 %load, 1 %ext = and i32 %add, 65535 @@ -169,14 +169,14 @@ define void @bfe_u32_zext_in_reg_i16_offset_8(i32 addrspace(1)* %out, i32 addrsp ; SI: v_and_b32_e32 {{v[0-9]+}}, 1, {{v[0-9]+}} ; SI: s_endpgm ; EG: AND_INT T{{[0-9]\.[XYZW]}}, T{{[0-9]\.[XYZW]}}, 1, -define void @bfe_u32_test_1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %x, i32 0, i32 1) store i32 %bfe, i32 addrspace(1)* %out, align 4 ret void } -define void @bfe_u32_test_2(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_2(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 0, i32 8) @@ -184,7 +184,7 @@ define void @bfe_u32_test_2(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ret void } -define void @bfe_u32_test_3(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_3(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 0, i32 1) @@ -199,7 +199,7 @@ define void @bfe_u32_test_3(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_mov_b32_e32 [[VREG:v[0-9]+]], 0 ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm -define void @bfe_u32_test_4(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_4(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %shr = lshr i32 %shl, 31 @@ -214,7 +214,7 @@ define void @bfe_u32_test_4(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI-NOT: shr ; SI: v_bfe_i32 {{v[0-9]+}}, {{v[0-9]+}}, 0, 1 ; SI: s_endpgm -define void @bfe_u32_test_5(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_5(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %shr = ashr i32 %shl, 31 @@ -227,7 +227,7 @@ define void @bfe_u32_test_5(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_lshlrev_b32_e32 v{{[0-9]+}}, 31, v{{[0-9]+}} ; SI: v_lshrrev_b32_e32 v{{[0-9]+}}, 1, v{{[0-9]+}} ; SI: s_endpgm -define void @bfe_u32_test_6(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_6(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 1, i32 31) @@ -239,7 +239,7 @@ define void @bfe_u32_test_6(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_lshlrev_b32_e32 v{{[0-9]+}}, 31, v{{[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 0, i32 31) @@ -252,7 +252,7 @@ define void @bfe_u32_test_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_and_b32_e32 {{v[0-9]+}}, 1, {{v[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = shl i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 31, i32 1) @@ -265,7 +265,7 @@ define void @bfe_u32_test_8(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_lshrrev_b32_e32 v{{[0-9]+}}, 31, v{{[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_9(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_9(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %x, i32 31, i32 1) store i32 %bfe, i32 addrspace(1)* %out, align 4 @@ -277,7 +277,7 @@ define void @bfe_u32_test_9(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounw ; SI: v_lshrrev_b32_e32 v{{[0-9]+}}, 1, v{{[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_10(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_10(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %x, i32 1, i32 31) store i32 %bfe, i32 addrspace(1)* %out, align 4 @@ -289,7 +289,7 @@ define void @bfe_u32_test_10(i32 addrspace(1)* %out, i32 addrspace(1)* %in) noun ; SI: v_lshrrev_b32_e32 v{{[0-9]+}}, 8, v{{[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_11(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_11(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %x, i32 8, i32 24) store i32 %bfe, i32 addrspace(1)* %out, align 4 @@ -301,7 +301,7 @@ define void @bfe_u32_test_11(i32 addrspace(1)* %out, i32 addrspace(1)* %in) noun ; SI: v_lshrrev_b32_e32 v{{[0-9]+}}, 24, v{{[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_12(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_12(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %x, i32 24, i32 8) store i32 %bfe, i32 addrspace(1)* %out, align 4 @@ -312,7 +312,7 @@ define void @bfe_u32_test_12(i32 addrspace(1)* %out, i32 addrspace(1)* %in) noun ; V_ASHRREV_U32_e32 {{v[0-9]+}}, 31, {{v[0-9]+}} ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_13(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_13(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = ashr i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 31, i32 1) @@ -323,7 +323,7 @@ define void @bfe_u32_test_13(i32 addrspace(1)* %out, i32 addrspace(1)* %in) noun ; SI-NOT: lshr ; SI-NOT: {{[^@]}}bfe ; SI: s_endpgm -define void @bfe_u32_test_14(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { +define amdgpu_kernel void @bfe_u32_test_14(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind { %x = load i32, i32 addrspace(1)* %in, align 4 %shl = lshr i32 %x, 31 %bfe = call i32 @llvm.AMDGPU.bfe.u32(i32 %shl, i32 31, i32 1) @@ -336,7 +336,7 @@ define void @bfe_u32_test_14(i32 addrspace(1)* %out, i32 addrspace(1)* %in) noun ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_0(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_0(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 0, i32 0, i32 0) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -348,7 +348,7 @@ define void @bfe_u32_constant_fold_test_0(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_1(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_1(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 12334, i32 0, i32 0) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -360,7 +360,7 @@ define void @bfe_u32_constant_fold_test_1(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_2(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_2(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 0, i32 0, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -372,7 +372,7 @@ define void @bfe_u32_constant_fold_test_2(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_3(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_3(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 1, i32 0, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -384,7 +384,7 @@ define void @bfe_u32_constant_fold_test_3(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_4(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_4(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 4294967295, i32 0, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -396,7 +396,7 @@ define void @bfe_u32_constant_fold_test_4(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_5(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_5(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 128, i32 7, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -408,7 +408,7 @@ define void @bfe_u32_constant_fold_test_5(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_6(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_6(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 128, i32 0, i32 8) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -420,7 +420,7 @@ define void @bfe_u32_constant_fold_test_6(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_7(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_7(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 127, i32 0, i32 8) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -432,7 +432,7 @@ define void @bfe_u32_constant_fold_test_7(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_8(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_8(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 127, i32 6, i32 8) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -444,7 +444,7 @@ define void @bfe_u32_constant_fold_test_8(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_9(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_9(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 65536, i32 16, i32 8) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -456,7 +456,7 @@ define void @bfe_u32_constant_fold_test_9(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_10(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_10(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 65535, i32 16, i32 16) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -468,7 +468,7 @@ define void @bfe_u32_constant_fold_test_10(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_11(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_11(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 160, i32 4, i32 4) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -480,7 +480,7 @@ define void @bfe_u32_constant_fold_test_11(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_12(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_12(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 160, i32 31, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -492,7 +492,7 @@ define void @bfe_u32_constant_fold_test_12(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_13(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_13(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 131070, i32 16, i32 16) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -504,7 +504,7 @@ define void @bfe_u32_constant_fold_test_13(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_14(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_14(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 160, i32 2, i32 30) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -516,7 +516,7 @@ define void @bfe_u32_constant_fold_test_14(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_15(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_15(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 160, i32 4, i32 28) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -528,7 +528,7 @@ define void @bfe_u32_constant_fold_test_15(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_16(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_16(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 4294967295, i32 1, i32 7) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -540,7 +540,7 @@ define void @bfe_u32_constant_fold_test_16(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_17(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_17(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 255, i32 1, i32 31) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -552,7 +552,7 @@ define void @bfe_u32_constant_fold_test_17(i32 addrspace(1)* %out) nounwind { ; SI: buffer_store_dword [[VREG]], ; SI: s_endpgm ; EG-NOT: BFE -define void @bfe_u32_constant_fold_test_18(i32 addrspace(1)* %out) nounwind { +define amdgpu_kernel void @bfe_u32_constant_fold_test_18(i32 addrspace(1)* %out) nounwind { %bfe_u32 = call i32 @llvm.AMDGPU.bfe.u32(i32 255, i32 31, i32 1) nounwind readnone store i32 %bfe_u32, i32 addrspace(1)* %out, align 4 ret void @@ -569,7 +569,7 @@ define void @bfe_u32_constant_fold_test_18(i32 addrspace(1)* %out) nounwind { ; SI-DAG: buffer_store_dword [[AND]] ; SI-DAG: buffer_store_dword [[BFE]] ; SI: s_endpgm -define void @simplify_bfe_u32_multi_use_arg(i32 addrspace(1)* %out0, +define amdgpu_kernel void @simplify_bfe_u32_multi_use_arg(i32 addrspace(1)* %out0, i32 addrspace(1)* %out1, i32 addrspace(1)* %in) nounwind { %src = load i32, i32 addrspace(1)* %in, align 4 @@ -583,7 +583,7 @@ define void @simplify_bfe_u32_multi_use_arg(i32 addrspace(1)* %out0, ; FUNC-LABEL: {{^}}lshr_and: ; SI: s_bfe_u32 {{s[0-9]+}}, {{s[0-9]+}}, 0x30006 ; SI: buffer_store_dword -define void @lshr_and(i32 addrspace(1)* %out, i32 %a) nounwind { +define amdgpu_kernel void @lshr_and(i32 addrspace(1)* %out, i32 %a) nounwind { %b = lshr i32 %a, 6 %c = and i32 %b, 7 store i32 %c, i32 addrspace(1)* %out, align 8 @@ -593,7 +593,7 @@ define void @lshr_and(i32 addrspace(1)* %out, i32 %a) nounwind { ; FUNC-LABEL: {{^}}v_lshr_and: ; SI: v_bfe_u32 {{v[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}, 3 ; SI: buffer_store_dword -define void @v_lshr_and(i32 addrspace(1)* %out, i32 %a, i32 %b) nounwind { +define amdgpu_kernel void @v_lshr_and(i32 addrspace(1)* %out, i32 %a, i32 %b) nounwind { %c = lshr i32 %a, %b %d = and i32 %c, 7 store i32 %d, i32 addrspace(1)* %out, align 8 @@ -603,7 +603,7 @@ define void @v_lshr_and(i32 addrspace(1)* %out, i32 %a, i32 %b) nounwind { ; FUNC-LABEL: {{^}}and_lshr: ; SI: s_bfe_u32 {{s[0-9]+}}, {{s[0-9]+}}, 0x30006 ; SI: buffer_store_dword -define void @and_lshr(i32 addrspace(1)* %out, i32 %a) nounwind { +define amdgpu_kernel void @and_lshr(i32 addrspace(1)* %out, i32 %a) nounwind { %b = and i32 %a, 448 %c = lshr i32 %b, 6 store i32 %c, i32 addrspace(1)* %out, align 8 @@ -613,7 +613,7 @@ define void @and_lshr(i32 addrspace(1)* %out, i32 %a) nounwind { ; FUNC-LABEL: {{^}}and_lshr2: ; SI: s_bfe_u32 {{s[0-9]+}}, {{s[0-9]+}}, 0x30006 ; SI: buffer_store_dword -define void @and_lshr2(i32 addrspace(1)* %out, i32 %a) nounwind { +define amdgpu_kernel void @and_lshr2(i32 addrspace(1)* %out, i32 %a) nounwind { %b = and i32 %a, 511 %c = lshr i32 %b, 6 store i32 %c, i32 addrspace(1)* %out, align 8 @@ -623,7 +623,7 @@ define void @and_lshr2(i32 addrspace(1)* %out, i32 %a) nounwind { ; FUNC-LABEL: {{^}}shl_lshr: ; SI: s_bfe_u32 {{s[0-9]+}}, {{s[0-9]+}}, 0x150002 ; SI: buffer_store_dword -define void @shl_lshr(i32 addrspace(1)* %out, i32 %a) nounwind { +define amdgpu_kernel void @shl_lshr(i32 addrspace(1)* %out, i32 %a) nounwind { %b = shl i32 %a, 9 %c = lshr i32 %b, 11 store i32 %c, i32 addrspace(1)* %out, align 8 |