AMDGPU/GlobalISel: legalize and select 32-bit G_ASHR

Reviewers: arsenm, nhaehnle

Subscribers: kzhuravl, wdng, yaxunl, rovka, kristof.beyls, dstuttard, tpr, llvm-commits, t-tye

Differential Revision: https://reviews.llvm.org/D48196

llvm-svn: 335318

2 files changed, 108 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-ashr.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-ashr.mir
new file mode 100644
index 00000000000..86692c3e1d6
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-ashr.mir
@@ -0,0 +1,86 @@
+# RUN: llc -march=amdgcn -mcpu=tahiti -run-pass=instruction-select -verify-machineinstrs -global-isel %s -o - | FileCheck %s -check-prefixes=GCN,SI
+# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=instruction-select -verify-machineinstrs -global-isel %s -o - | FileCheck %s -check-prefixes=GCN,VI
+
+--- |
+  define void @ashr(i32 addrspace(1)* %global0) {ret void}
+...
+---
+
+name:            ashr
+legalized:       true
+regBankSelected: true
+
+# GCN-LABEL: name: ashr
+body: |
+  bb.0:
+    liveins: $sgpr0, $sgpr1, $vgpr0, $vgpr3_vgpr4
+    ; GCN: [[SGPR0:%[0-9]+]]:sreg_32 = COPY $sgpr0
+    ; GCN: [[SGPR1:%[0-9]+]]:sreg_32 = COPY $sgpr1
+    ; GCN: [[VGPR0:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+    %0:sgpr(s32) = COPY $sgpr0
+    %1:sgpr(s32) = COPY $sgpr1
+    %2:vgpr(s32) = COPY $vgpr0
+    %3:vgpr(s64) = COPY $vgpr3_vgpr4
+
+    ; GCN: [[C1:%[0-9]+]]:sreg_32_xm0 = S_MOV_B32 1
+    ; GCN: [[C4096:%[0-9]+]]:sreg_32_xm0 = S_MOV_B32 4096
+    %4:sgpr(s32) = G_CONSTANT i32 1
+    %5:sgpr(s32) = G_CONSTANT i32 4096
+
+    ; ashr ss
+    ; GCN: [[SS:%[0-9]+]]:sreg_32 = S_ASHR_I32 [[SGPR0]], [[SGPR1]]
+    %6:sgpr(s32) = G_ASHR %0, %1
+
+    ; ashr si
+    ; GCN: [[SI:%[0-9]+]]:sreg_32 = S_ASHR_I32 [[SS]], [[C1]]
+    %7:sgpr(s32) = G_ASHR %6, %4
+
+    ; ashr is
+    ; GCN: [[IS:%[0-9]+]]:sreg_32 = S_ASHR_I32 [[C1]], [[SI]]
+    %8:sgpr(s32) = G_ASHR %4, %7
+
+    ; ashr sc
+    ; GCN: [[SC:%[0-9]+]]:sreg_32 = S_ASHR_I32 [[IS]], [[C4096]]
+    %9:sgpr(s32) = G_ASHR %8, %5
+
+    ; ashr cs
+    ; GCN: [[CS:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[C4096]], [[SC]]
+    %10:sgpr(s32) = G_ASHR %5, %9
+
+    ; ashr vs
+    ; GCN: [[VS:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e32 [[CS]], [[VGPR0]]
+    %11:vgpr(s32) = G_ASHR %2, %10
+
+    ; ashr sv
+    ; SI: [[SV:%[0-9]+]]:vgpr_32 = V_ASHR_I32_e32 [[CS]], [[VS]]
+    ; VI: [[SV:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e64 [[VS]], [[CS]]
+    %12:vgpr(s32) = G_ASHR %10, %11
+
+    ; ashr vv
+    ; SI: [[VV:%[0-9]+]]:vgpr_32 = V_ASHR_I32_e32 [[SV]], [[VGPR0]]
+    ; VI: [[VV:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e32 [[VGPR0]], [[SV]]
+    %13:vgpr(s32) = G_ASHR %12, %2
+
+    ; ashr iv
+    ; SI: [[IV:%[0-9]+]]:vgpr_32 = V_ASHR_I32_e32 [[C1]], [[VV]]
+    ; VI: [[IV:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e64 [[VV]], [[C1]]
+    %14:vgpr(s32) = G_ASHR %4, %13
+
+    ; ashr vi
+    ; GCN: [[VI:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e32 [[C1]], [[IV]]
+    %15:vgpr(s32) = G_ASHR %14, %4
+
+    ; ashr cv
+    ; SI: [[CV:%[0-9]+]]:vgpr_32 = V_ASHR_I32_e32 [[C4096]], [[VI]]
+    ; VI: [[CV:%[0-9]+]]:vgpr_32 = V_ASHRREV_I32_e64 [[VI]], [[C4096]]
+    %16:vgpr(s32) = G_ASHR %5, %15
+
+    ; ashr vc
+    ; GCN: [[VC:%[-1-9]+]]:vgpr_32 = V_ASHRREV_I32_e32 [[C4096]], [[CV]]
+    %17:vgpr(s32) = G_ASHR %16, %5
+
+
+    G_STORE %17, %3 :: (store 4 into %ir.global0)
+
+...
+---
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-ashr.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-ashr.mir
new file mode 100644
index 00000000000..71a9de8e6bf
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-ashr.mir
@@ -0,0 +1,22 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -O0 -run-pass=legalizer -global-isel %s -o - | FileCheck %s
+
+---
+name:            test_ashr
+registers:
+  - { id: 0, class: _ }
+  - { id: 1, class: _ }
+  - { id: 2, class: _ }
+body: |
+  bb.0.entry:
+    liveins: $vgpr0, $vgpr1
+
+    ; CHECK-LABEL: name: test_ashr
+    ; CHECK: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+    ; CHECK: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+    ; CHECK: [[SHL:%[0-9]+]]:_(s32) = G_ASHR [[COPY]], [[COPY1]]
+    %0(s32) = COPY $vgpr0
+    %1(s32) = COPY $vgpr1
+    %2(s32) = G_ASHR %0, %1
+    $vgpr0 = COPY %2
+...