From 25e60f645ac8bf09e00bfff95c9d20d63616d687 Mon Sep 17 00:00:00 2001 From: Matt Arsenault Date: Sat, 27 Feb 2021 14:41:45 -0500 Subject: [PATCH] AMDGPU/GlobalISel: Add subtarget to a test SelectionDAG forces us to have a weird ABI for 16-bit values without legal 16-bit operations, but currently GlobalISel bypasses this and sometimes ends up using the gfx8+ ABI in some contexts. Make sure we're testing the normal ABI to avoid a test change in a future patch. --- .../CodeGen/AMDGPU/GlobalISel/function-returns.ll | 274 +++++++++------------ 1 file changed, 119 insertions(+), 155 deletions(-) diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/function-returns.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/function-returns.ll index 2a9bd6b..e6571b9 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/function-returns.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/function-returns.ll @@ -1,5 +1,7 @@ ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py -; RUN: llc -global-isel -global-isel-abort=0 -stop-after=irtranslator -mtriple=amdgcn-mesa-mesa3d -verify-machineinstrs -o - %s | FileCheck -enable-var-scope %s +; RUN: llc -global-isel -global-isel-abort=0 -stop-after=irtranslator -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -verify-machineinstrs -o - %s | FileCheck -enable-var-scope %s + +; FIXME: Also test with a pre-gfx8 target. define i1 @i1_func_void() #0 { ; CHECK-LABEL: name: i1_func_void @@ -813,13 +815,9 @@ define <2 x i16> @v2i16_func_void() #0 { ; CHECK: [[COPY:%[0-9]+]]:sgpr_64 = COPY $sgpr30_sgpr31 ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<2 x s16>) = G_LOAD [[DEF]](p1) :: (load 4 from `<2 x i16> addrspace(1)* undef`, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD]](<2 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) + ; CHECK: $vgpr0 = COPY [[LOAD]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0 %val = load <2 x i16>, <2 x i16> addrspace(1)* undef ret <2 x i16> %val } @@ -831,13 +829,9 @@ define <2 x half> @v2f16_func_void() #0 { ; CHECK: [[COPY:%[0-9]+]]:sgpr_64 = COPY $sgpr30_sgpr31 ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<2 x s16>) = G_LOAD [[DEF]](p1) :: (load 4 from `<2 x half> addrspace(1)* undef`, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD]](<2 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) + ; CHECK: $vgpr0 = COPY [[LOAD]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0 %val = load <2 x half>, <2 x half> addrspace(1)* undef ret <2 x half> %val } @@ -849,15 +843,13 @@ define <3 x i16> @v3i16_func_void() #0 { ; CHECK: [[COPY:%[0-9]+]]:sgpr_64 = COPY $sgpr30_sgpr31 ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<3 x s16>) = G_LOAD [[DEF]](p1) :: (load 6 from `<3 x i16> addrspace(1)* undef`, align 8, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD]](<3 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) + ; CHECK: [[DEF1:%[0-9]+]]:_(<3 x s16>) = G_IMPLICIT_DEF + ; CHECK: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[LOAD]](<3 x s16>), [[DEF1]](<3 x s16>) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[CONCAT_VECTORS]](<6 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 %val = load <3 x i16>, <3 x i16> addrspace(1)* undef ret <3 x i16> %val } @@ -869,17 +861,11 @@ define <4 x i16> @v4i16_func_void() #0 { ; CHECK: [[COPY:%[0-9]+]]:sgpr_64 = COPY $sgpr30_sgpr31 ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<4 x s16>) = G_LOAD [[DEF]](p1) :: (load 8 from `<4 x i16> addrspace(1)* undef`, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16), [[UV3:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD]](<4 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[LOAD]](<4 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 %val = load <4 x i16>, <4 x i16> addrspace(1)* undef ret <4 x i16> %val } @@ -891,17 +877,11 @@ define <4 x half> @v4f16_func_void() #0 { ; CHECK: [[COPY:%[0-9]+]]:sgpr_64 = COPY $sgpr30_sgpr31 ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<4 x s16>) = G_LOAD [[DEF]](p1) :: (load 8 from `<4 x half> addrspace(1)* undef`, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16), [[UV3:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD]](<4 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[LOAD]](<4 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 %val = load <4 x half>, <4 x half> addrspace(1)* undef ret <4 x half> %val } @@ -914,19 +894,14 @@ define <5 x i16> @v5i16_func_void() #0 { ; CHECK: [[DEF:%[0-9]+]]:_(p4) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(p1) = G_LOAD [[DEF]](p4) :: (volatile load 8 from `<5 x i16> addrspace(1)* addrspace(4)* undef`, addrspace 4) ; CHECK: [[LOAD1:%[0-9]+]]:_(<5 x s16>) = G_LOAD [[LOAD]](p1) :: (load 10 from %ir.ptr, align 16, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16), [[UV3:%[0-9]+]]:_(s16), [[UV4:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD1]](<5 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s16) - ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[UV4]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) - ; CHECK: $vgpr4 = COPY [[ANYEXT4]](s32) + ; CHECK: [[DEF1:%[0-9]+]]:_(<5 x s16>) = G_IMPLICIT_DEF + ; CHECK: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[LOAD1]](<5 x s16>), [[DEF1]](<5 x s16>) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>), [[UV4:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[CONCAT_VECTORS]](<10 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) + ; CHECK: $vgpr2 = COPY [[UV2]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3, implicit $vgpr4 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 %ptr = load volatile <5 x i16> addrspace(1)*, <5 x i16> addrspace(1)* addrspace(4)* undef %val = load <5 x i16>, <5 x i16> addrspace(1)* %ptr ret <5 x i16> %val @@ -940,25 +915,13 @@ define <8 x i16> @v8i16_func_void() #0 { ; CHECK: [[DEF:%[0-9]+]]:_(p4) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(p1) = G_LOAD [[DEF]](p4) :: (volatile load 8 from `<8 x i16> addrspace(1)* addrspace(4)* undef`, addrspace 4) ; CHECK: [[LOAD1:%[0-9]+]]:_(<8 x s16>) = G_LOAD [[LOAD]](p1) :: (load 16 from %ir.ptr, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16), [[UV3:%[0-9]+]]:_(s16), [[UV4:%[0-9]+]]:_(s16), [[UV5:%[0-9]+]]:_(s16), [[UV6:%[0-9]+]]:_(s16), [[UV7:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD1]](<8 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s16) - ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[UV4]](s16) - ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s32) = G_ANYEXT [[UV5]](s16) - ; CHECK: [[ANYEXT6:%[0-9]+]]:_(s32) = G_ANYEXT [[UV6]](s16) - ; CHECK: [[ANYEXT7:%[0-9]+]]:_(s32) = G_ANYEXT [[UV7]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) - ; CHECK: $vgpr4 = COPY [[ANYEXT4]](s32) - ; CHECK: $vgpr5 = COPY [[ANYEXT5]](s32) - ; CHECK: $vgpr6 = COPY [[ANYEXT6]](s32) - ; CHECK: $vgpr7 = COPY [[ANYEXT7]](s32) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[LOAD1]](<8 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) + ; CHECK: $vgpr2 = COPY [[UV2]](<2 x s16>) + ; CHECK: $vgpr3 = COPY [[UV3]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3, implicit $vgpr4, implicit $vgpr5, implicit $vgpr6, implicit $vgpr7 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 %ptr = load volatile <8 x i16> addrspace(1)*, <8 x i16> addrspace(1)* addrspace(4)* undef %val = load <8 x i16>, <8 x i16> addrspace(1)* %ptr ret <8 x i16> %val @@ -972,41 +935,17 @@ define <16 x i16> @v16i16_func_void() #0 { ; CHECK: [[DEF:%[0-9]+]]:_(p4) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(p1) = G_LOAD [[DEF]](p4) :: (volatile load 8 from `<16 x i16> addrspace(1)* addrspace(4)* undef`, addrspace 4) ; CHECK: [[LOAD1:%[0-9]+]]:_(<16 x s16>) = G_LOAD [[LOAD]](p1) :: (load 32 from %ir.ptr, addrspace 1) - ; CHECK: [[UV:%[0-9]+]]:_(s16), [[UV1:%[0-9]+]]:_(s16), [[UV2:%[0-9]+]]:_(s16), [[UV3:%[0-9]+]]:_(s16), [[UV4:%[0-9]+]]:_(s16), [[UV5:%[0-9]+]]:_(s16), [[UV6:%[0-9]+]]:_(s16), [[UV7:%[0-9]+]]:_(s16), [[UV8:%[0-9]+]]:_(s16), [[UV9:%[0-9]+]]:_(s16), [[UV10:%[0-9]+]]:_(s16), [[UV11:%[0-9]+]]:_(s16), [[UV12:%[0-9]+]]:_(s16), [[UV13:%[0-9]+]]:_(s16), [[UV14:%[0-9]+]]:_(s16), [[UV15:%[0-9]+]]:_(s16) = G_UNMERGE_VALUES [[LOAD1]](<16 x s16>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s16) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s16) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s16) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s16) - ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[UV4]](s16) - ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s32) = G_ANYEXT [[UV5]](s16) - ; CHECK: [[ANYEXT6:%[0-9]+]]:_(s32) = G_ANYEXT [[UV6]](s16) - ; CHECK: [[ANYEXT7:%[0-9]+]]:_(s32) = G_ANYEXT [[UV7]](s16) - ; CHECK: [[ANYEXT8:%[0-9]+]]:_(s32) = G_ANYEXT [[UV8]](s16) - ; CHECK: [[ANYEXT9:%[0-9]+]]:_(s32) = G_ANYEXT [[UV9]](s16) - ; CHECK: [[ANYEXT10:%[0-9]+]]:_(s32) = G_ANYEXT [[UV10]](s16) - ; CHECK: [[ANYEXT11:%[0-9]+]]:_(s32) = G_ANYEXT [[UV11]](s16) - ; CHECK: [[ANYEXT12:%[0-9]+]]:_(s32) = G_ANYEXT [[UV12]](s16) - ; CHECK: [[ANYEXT13:%[0-9]+]]:_(s32) = G_ANYEXT [[UV13]](s16) - ; CHECK: [[ANYEXT14:%[0-9]+]]:_(s32) = G_ANYEXT [[UV14]](s16) - ; CHECK: [[ANYEXT15:%[0-9]+]]:_(s32) = G_ANYEXT [[UV15]](s16) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) - ; CHECK: $vgpr4 = COPY [[ANYEXT4]](s32) - ; CHECK: $vgpr5 = COPY [[ANYEXT5]](s32) - ; CHECK: $vgpr6 = COPY [[ANYEXT6]](s32) - ; CHECK: $vgpr7 = COPY [[ANYEXT7]](s32) - ; CHECK: $vgpr8 = COPY [[ANYEXT8]](s32) - ; CHECK: $vgpr9 = COPY [[ANYEXT9]](s32) - ; CHECK: $vgpr10 = COPY [[ANYEXT10]](s32) - ; CHECK: $vgpr11 = COPY [[ANYEXT11]](s32) - ; CHECK: $vgpr12 = COPY [[ANYEXT12]](s32) - ; CHECK: $vgpr13 = COPY [[ANYEXT13]](s32) - ; CHECK: $vgpr14 = COPY [[ANYEXT14]](s32) - ; CHECK: $vgpr15 = COPY [[ANYEXT15]](s32) + ; CHECK: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>), [[UV4:%[0-9]+]]:_(<2 x s16>), [[UV5:%[0-9]+]]:_(<2 x s16>), [[UV6:%[0-9]+]]:_(<2 x s16>), [[UV7:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[LOAD1]](<16 x s16>) + ; CHECK: $vgpr0 = COPY [[UV]](<2 x s16>) + ; CHECK: $vgpr1 = COPY [[UV1]](<2 x s16>) + ; CHECK: $vgpr2 = COPY [[UV2]](<2 x s16>) + ; CHECK: $vgpr3 = COPY [[UV3]](<2 x s16>) + ; CHECK: $vgpr4 = COPY [[UV4]](<2 x s16>) + ; CHECK: $vgpr5 = COPY [[UV5]](<2 x s16>) + ; CHECK: $vgpr6 = COPY [[UV6]](<2 x s16>) + ; CHECK: $vgpr7 = COPY [[UV7]](<2 x s16>) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] - ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3, implicit $vgpr4, implicit $vgpr5, implicit $vgpr6, implicit $vgpr7, implicit $vgpr8, implicit $vgpr9, implicit $vgpr10, implicit $vgpr11, implicit $vgpr12, implicit $vgpr13, implicit $vgpr14, implicit $vgpr15 + ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3, implicit $vgpr4, implicit $vgpr5, implicit $vgpr6, implicit $vgpr7 %ptr = load volatile <16 x i16> addrspace(1)*, <16 x i16> addrspace(1)* addrspace(4)* undef %val = load <16 x i16>, <16 x i16> addrspace(1)* %ptr ret <16 x i16> %val @@ -1021,38 +960,54 @@ define <16 x i8> @v16i8_func_void() #0 { ; CHECK: [[LOAD:%[0-9]+]]:_(p1) = G_LOAD [[DEF]](p4) :: (volatile load 8 from `<16 x i8> addrspace(1)* addrspace(4)* undef`, addrspace 4) ; CHECK: [[LOAD1:%[0-9]+]]:_(<16 x s8>) = G_LOAD [[LOAD]](p1) :: (load 16 from %ir.ptr, addrspace 1) ; CHECK: [[UV:%[0-9]+]]:_(s8), [[UV1:%[0-9]+]]:_(s8), [[UV2:%[0-9]+]]:_(s8), [[UV3:%[0-9]+]]:_(s8), [[UV4:%[0-9]+]]:_(s8), [[UV5:%[0-9]+]]:_(s8), [[UV6:%[0-9]+]]:_(s8), [[UV7:%[0-9]+]]:_(s8), [[UV8:%[0-9]+]]:_(s8), [[UV9:%[0-9]+]]:_(s8), [[UV10:%[0-9]+]]:_(s8), [[UV11:%[0-9]+]]:_(s8), [[UV12:%[0-9]+]]:_(s8), [[UV13:%[0-9]+]]:_(s8), [[UV14:%[0-9]+]]:_(s8), [[UV15:%[0-9]+]]:_(s8) = G_UNMERGE_VALUES [[LOAD1]](<16 x s8>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s8) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s8) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s8) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s8) - ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[UV4]](s8) - ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s32) = G_ANYEXT [[UV5]](s8) - ; CHECK: [[ANYEXT6:%[0-9]+]]:_(s32) = G_ANYEXT [[UV6]](s8) - ; CHECK: [[ANYEXT7:%[0-9]+]]:_(s32) = G_ANYEXT [[UV7]](s8) - ; CHECK: [[ANYEXT8:%[0-9]+]]:_(s32) = G_ANYEXT [[UV8]](s8) - ; CHECK: [[ANYEXT9:%[0-9]+]]:_(s32) = G_ANYEXT [[UV9]](s8) - ; CHECK: [[ANYEXT10:%[0-9]+]]:_(s32) = G_ANYEXT [[UV10]](s8) - ; CHECK: [[ANYEXT11:%[0-9]+]]:_(s32) = G_ANYEXT [[UV11]](s8) - ; CHECK: [[ANYEXT12:%[0-9]+]]:_(s32) = G_ANYEXT [[UV12]](s8) - ; CHECK: [[ANYEXT13:%[0-9]+]]:_(s32) = G_ANYEXT [[UV13]](s8) - ; CHECK: [[ANYEXT14:%[0-9]+]]:_(s32) = G_ANYEXT [[UV14]](s8) - ; CHECK: [[ANYEXT15:%[0-9]+]]:_(s32) = G_ANYEXT [[UV15]](s8) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) - ; CHECK: $vgpr4 = COPY [[ANYEXT4]](s32) - ; CHECK: $vgpr5 = COPY [[ANYEXT5]](s32) - ; CHECK: $vgpr6 = COPY [[ANYEXT6]](s32) - ; CHECK: $vgpr7 = COPY [[ANYEXT7]](s32) - ; CHECK: $vgpr8 = COPY [[ANYEXT8]](s32) - ; CHECK: $vgpr9 = COPY [[ANYEXT9]](s32) - ; CHECK: $vgpr10 = COPY [[ANYEXT10]](s32) - ; CHECK: $vgpr11 = COPY [[ANYEXT11]](s32) - ; CHECK: $vgpr12 = COPY [[ANYEXT12]](s32) - ; CHECK: $vgpr13 = COPY [[ANYEXT13]](s32) - ; CHECK: $vgpr14 = COPY [[ANYEXT14]](s32) - ; CHECK: $vgpr15 = COPY [[ANYEXT15]](s32) + ; CHECK: [[ANYEXT:%[0-9]+]]:_(s16) = G_ANYEXT [[UV]](s8) + ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s16) = G_ANYEXT [[UV1]](s8) + ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s16) = G_ANYEXT [[UV2]](s8) + ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s16) = G_ANYEXT [[UV3]](s8) + ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s16) = G_ANYEXT [[UV4]](s8) + ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s16) = G_ANYEXT [[UV5]](s8) + ; CHECK: [[ANYEXT6:%[0-9]+]]:_(s16) = G_ANYEXT [[UV6]](s8) + ; CHECK: [[ANYEXT7:%[0-9]+]]:_(s16) = G_ANYEXT [[UV7]](s8) + ; CHECK: [[ANYEXT8:%[0-9]+]]:_(s16) = G_ANYEXT [[UV8]](s8) + ; CHECK: [[ANYEXT9:%[0-9]+]]:_(s16) = G_ANYEXT [[UV9]](s8) + ; CHECK: [[ANYEXT10:%[0-9]+]]:_(s16) = G_ANYEXT [[UV10]](s8) + ; CHECK: [[ANYEXT11:%[0-9]+]]:_(s16) = G_ANYEXT [[UV11]](s8) + ; CHECK: [[ANYEXT12:%[0-9]+]]:_(s16) = G_ANYEXT [[UV12]](s8) + ; CHECK: [[ANYEXT13:%[0-9]+]]:_(s16) = G_ANYEXT [[UV13]](s8) + ; CHECK: [[ANYEXT14:%[0-9]+]]:_(s16) = G_ANYEXT [[UV14]](s8) + ; CHECK: [[ANYEXT15:%[0-9]+]]:_(s16) = G_ANYEXT [[UV15]](s8) + ; CHECK: [[ANYEXT16:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT]](s16) + ; CHECK: $vgpr0 = COPY [[ANYEXT16]](s32) + ; CHECK: [[ANYEXT17:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT1]](s16) + ; CHECK: $vgpr1 = COPY [[ANYEXT17]](s32) + ; CHECK: [[ANYEXT18:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT2]](s16) + ; CHECK: $vgpr2 = COPY [[ANYEXT18]](s32) + ; CHECK: [[ANYEXT19:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT3]](s16) + ; CHECK: $vgpr3 = COPY [[ANYEXT19]](s32) + ; CHECK: [[ANYEXT20:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT4]](s16) + ; CHECK: $vgpr4 = COPY [[ANYEXT20]](s32) + ; CHECK: [[ANYEXT21:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT5]](s16) + ; CHECK: $vgpr5 = COPY [[ANYEXT21]](s32) + ; CHECK: [[ANYEXT22:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT6]](s16) + ; CHECK: $vgpr6 = COPY [[ANYEXT22]](s32) + ; CHECK: [[ANYEXT23:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT7]](s16) + ; CHECK: $vgpr7 = COPY [[ANYEXT23]](s32) + ; CHECK: [[ANYEXT24:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT8]](s16) + ; CHECK: $vgpr8 = COPY [[ANYEXT24]](s32) + ; CHECK: [[ANYEXT25:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT9]](s16) + ; CHECK: $vgpr9 = COPY [[ANYEXT25]](s32) + ; CHECK: [[ANYEXT26:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT10]](s16) + ; CHECK: $vgpr10 = COPY [[ANYEXT26]](s32) + ; CHECK: [[ANYEXT27:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT11]](s16) + ; CHECK: $vgpr11 = COPY [[ANYEXT27]](s32) + ; CHECK: [[ANYEXT28:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT12]](s16) + ; CHECK: $vgpr12 = COPY [[ANYEXT28]](s32) + ; CHECK: [[ANYEXT29:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT13]](s16) + ; CHECK: $vgpr13 = COPY [[ANYEXT29]](s32) + ; CHECK: [[ANYEXT30:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT14]](s16) + ; CHECK: $vgpr14 = COPY [[ANYEXT30]](s32) + ; CHECK: [[ANYEXT31:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT15]](s16) + ; CHECK: $vgpr15 = COPY [[ANYEXT31]](s32) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3, implicit $vgpr4, implicit $vgpr5, implicit $vgpr6, implicit $vgpr7, implicit $vgpr8, implicit $vgpr9, implicit $vgpr10, implicit $vgpr11, implicit $vgpr12, implicit $vgpr13, implicit $vgpr14, implicit $vgpr15 %ptr = load volatile <16 x i8> addrspace(1)*, <16 x i8> addrspace(1)* addrspace(4)* undef @@ -1068,10 +1023,12 @@ define <2 x i8> @v2i8_func_void() #0 { ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<2 x s8>) = G_LOAD [[DEF]](p1) :: (load 2 from `<2 x i8> addrspace(1)* undef`, addrspace 1) ; CHECK: [[UV:%[0-9]+]]:_(s8), [[UV1:%[0-9]+]]:_(s8) = G_UNMERGE_VALUES [[LOAD]](<2 x s8>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s8) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s8) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) + ; CHECK: [[ANYEXT:%[0-9]+]]:_(s16) = G_ANYEXT [[UV]](s8) + ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s16) = G_ANYEXT [[UV1]](s8) + ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT]](s16) + ; CHECK: $vgpr0 = COPY [[ANYEXT2]](s32) + ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT1]](s16) + ; CHECK: $vgpr1 = COPY [[ANYEXT3]](s32) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1 %val = load <2 x i8>, <2 x i8> addrspace(1)* undef @@ -1086,12 +1043,15 @@ define <3 x i8> @v3i8_func_void() #0 { ; CHECK: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; CHECK: [[LOAD:%[0-9]+]]:_(<3 x s8>) = G_LOAD [[DEF]](p1) :: (load 3 from `<3 x i8> addrspace(1)* undef`, align 4, addrspace 1) ; CHECK: [[UV:%[0-9]+]]:_(s8), [[UV1:%[0-9]+]]:_(s8), [[UV2:%[0-9]+]]:_(s8) = G_UNMERGE_VALUES [[LOAD]](<3 x s8>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s8) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s8) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s8) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) + ; CHECK: [[ANYEXT:%[0-9]+]]:_(s16) = G_ANYEXT [[UV]](s8) + ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s16) = G_ANYEXT [[UV1]](s8) + ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s16) = G_ANYEXT [[UV2]](s8) + ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT]](s16) + ; CHECK: $vgpr0 = COPY [[ANYEXT3]](s32) + ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT1]](s16) + ; CHECK: $vgpr1 = COPY [[ANYEXT4]](s32) + ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT2]](s16) + ; CHECK: $vgpr2 = COPY [[ANYEXT5]](s32) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 %val = load <3 x i8>, <3 x i8> addrspace(1)* undef @@ -1107,14 +1067,18 @@ define <4 x i8> @v4i8_func_void() #0 { ; CHECK: [[LOAD:%[0-9]+]]:_(p1) = G_LOAD [[DEF]](p4) :: (volatile load 8 from `<4 x i8> addrspace(1)* addrspace(4)* undef`, addrspace 4) ; CHECK: [[LOAD1:%[0-9]+]]:_(<4 x s8>) = G_LOAD [[LOAD]](p1) :: (load 4 from %ir.ptr, addrspace 1) ; CHECK: [[UV:%[0-9]+]]:_(s8), [[UV1:%[0-9]+]]:_(s8), [[UV2:%[0-9]+]]:_(s8), [[UV3:%[0-9]+]]:_(s8) = G_UNMERGE_VALUES [[LOAD1]](<4 x s8>) - ; CHECK: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[UV]](s8) - ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[UV1]](s8) - ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[UV2]](s8) - ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[UV3]](s8) - ; CHECK: $vgpr0 = COPY [[ANYEXT]](s32) - ; CHECK: $vgpr1 = COPY [[ANYEXT1]](s32) - ; CHECK: $vgpr2 = COPY [[ANYEXT2]](s32) - ; CHECK: $vgpr3 = COPY [[ANYEXT3]](s32) + ; CHECK: [[ANYEXT:%[0-9]+]]:_(s16) = G_ANYEXT [[UV]](s8) + ; CHECK: [[ANYEXT1:%[0-9]+]]:_(s16) = G_ANYEXT [[UV1]](s8) + ; CHECK: [[ANYEXT2:%[0-9]+]]:_(s16) = G_ANYEXT [[UV2]](s8) + ; CHECK: [[ANYEXT3:%[0-9]+]]:_(s16) = G_ANYEXT [[UV3]](s8) + ; CHECK: [[ANYEXT4:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT]](s16) + ; CHECK: $vgpr0 = COPY [[ANYEXT4]](s32) + ; CHECK: [[ANYEXT5:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT1]](s16) + ; CHECK: $vgpr1 = COPY [[ANYEXT5]](s32) + ; CHECK: [[ANYEXT6:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT2]](s16) + ; CHECK: $vgpr2 = COPY [[ANYEXT6]](s32) + ; CHECK: [[ANYEXT7:%[0-9]+]]:_(s32) = G_ANYEXT [[ANYEXT3]](s16) + ; CHECK: $vgpr3 = COPY [[ANYEXT7]](s32) ; CHECK: [[COPY1:%[0-9]+]]:ccr_sgpr_64 = COPY [[COPY]] ; CHECK: S_SETPC_B64_return [[COPY1]], implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 %ptr = load volatile <4 x i8> addrspace(1)*, <4 x i8> addrspace(1)* addrspace(4)* undef -- 2.7.4