[AMDGPU][GlobalISel] Fix issue with copy_scc_vcc on gfx7 (#165355)

When selecting for G_AMDGPU_COPY_SCC_VCC, we use S_CMP_LG_U64 or
S_CMP_LG_U32 for wave64 and wave32 respectively. However, on gfx7 we do
not have the S_CMP_LG_U64 instruction. Work around this issue by using
S_OR_B64 instead.

Author: vangthao95

llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp

@@ -221,12 +221,22 @@ bool AMDGPUInstructionSelector::selectCOPY(MachineInstr &I) const {
 bool AMDGPUInstructionSelector::selectCOPY_SCC_VCC(MachineInstr &I) const {
   const DebugLoc &DL = I.getDebugLoc();
   MachineBasicBlock *BB = I.getParent();
+  Register VCCReg = I.getOperand(1).getReg();
+  MachineInstr *Cmp;
+
+  if (STI.getGeneration() >= AMDGPUSubtarget::VOLCANIC_ISLANDS) {
+    unsigned CmpOpc =
+        STI.isWave64() ? AMDGPU::S_CMP_LG_U64 : AMDGPU::S_CMP_LG_U32;
+    Cmp = BuildMI(*BB, &I, DL, TII.get(CmpOpc)).addReg(VCCReg).addImm(0);
+  } else {
+    // For gfx7 and earlier, S_CMP_LG_U64 doesn't exist, so we use S_OR_B64
+    // which sets SCC as a side effect.
+    Register DeadDst = MRI->createVirtualRegister(&AMDGPU::SReg_64RegClass);
+    Cmp = BuildMI(*BB, &I, DL, TII.get(AMDGPU::S_OR_B64), DeadDst)
+              .addReg(VCCReg)
+              .addReg(VCCReg);
+  }
 
-  unsigned CmpOpc =
-      STI.isWave64() ? AMDGPU::S_CMP_LG_U64 : AMDGPU::S_CMP_LG_U32;
-  MachineInstr *Cmp = BuildMI(*BB, &I, DL, TII.get(CmpOpc))
-                          .addReg(I.getOperand(1).getReg())
-                          .addImm(0);
   if (!constrainSelectedInstRegOperands(*Cmp, TII, TRI, RBI))
     return false;
 

llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-copy-scc-vcc.ll

@@ -0,0 +1,66 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx700 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX7 %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx803 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX8 %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1100 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s
+
+define amdgpu_kernel void @fcmp_uniform_select(float %a, i32 %b, i32 %c, ptr addrspace(1) %out) {
+; GFX7-LABEL: fcmp_uniform_select:
+; GFX7:       ; %bb.0:
+; GFX7-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dword s3, s[4:5], 0xb
+; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    v_cmp_eq_f32_e64 s[4:5], s6, 0
+; GFX7-NEXT:    s_or_b64 s[4:5], s[4:5], s[4:5]
+; GFX7-NEXT:    s_cselect_b32 s4, 1, 0
+; GFX7-NEXT:    s_and_b32 s4, s4, 1
+; GFX7-NEXT:    s_cmp_lg_u32 s4, 0
+; GFX7-NEXT:    s_cselect_b32 s3, s7, s3
+; GFX7-NEXT:    v_mov_b32_e32 v0, s3
+; GFX7-NEXT:    s_mov_b32 s3, 0xf000
+; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    s_endpgm
+;
+; GFX8-LABEL: fcmp_uniform_select:
+; GFX8:       ; %bb.0:
+; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX8-NEXT:    s_load_dword s6, s[4:5], 0x2c
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[4:5], 0x34
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    v_cmp_eq_f32_e64 s[4:5], s0, 0
+; GFX8-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX8-NEXT:    s_cselect_b32 s0, 1, 0
+; GFX8-NEXT:    s_and_b32 s0, s0, 1
+; GFX8-NEXT:    s_cmp_lg_u32 s0, 0
+; GFX8-NEXT:    s_cselect_b32 s0, s1, s6
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v2, s0
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
+; GFX8-NEXT:    flat_store_dword v[0:1], v2
+; GFX8-NEXT:    s_endpgm
+;
+; GFX11-LABEL: fcmp_uniform_select:
+; GFX11:       ; %bb.0:
+; GFX11-NEXT:    s_clause 0x2
+; GFX11-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX11-NEXT:    s_load_b32 s6, s[4:5], 0x2c
+; GFX11-NEXT:    s_load_b64 s[2:3], s[4:5], 0x34
+; GFX11-NEXT:    v_mov_b32_e32 v1, 0
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    v_cmp_eq_f32_e64 s0, s0, 0
+; GFX11-NEXT:    s_cmp_lg_u32 s0, 0
+; GFX11-NEXT:    s_cselect_b32 s0, 1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_and_b32 s0, s0, 1
+; GFX11-NEXT:    s_cmp_lg_u32 s0, 0
+; GFX11-NEXT:    s_cselect_b32 s0, s1, s6
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-NEXT:    v_mov_b32_e32 v0, s0
+; GFX11-NEXT:    global_store_b32 v1, v0, s[2:3]
+; GFX11-NEXT:    s_endpgm
+  %cmp = fcmp oeq float %a, 0.0
+  %sel = select i1 %cmp, i32 %b, i32 %c
+  store i32 %sel, ptr addrspace(1) %out
+  ret void
+}

llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-copy-scc-vcc.mir

@@ -0,0 +1,37 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn -mcpu=gfx700 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefixes=GFX7 %s
+# RUN: llc -mtriple=amdgcn -mcpu=gfx803 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefixes=GF8 %s
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefixes=GFX11 %s
+
+---
+name: test_copy_scc_vcc
+legalized: true
+regBankSelected: true
+tracksRegLiveness: true
+body: |
+  bb.0:
+    ; GFX7-LABEL: name: test_copy_scc_vcc
+    ; GFX7: [[DEF:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; GFX7-NEXT: [[S_OR_B64_:%[0-9]+]]:sreg_64 = S_OR_B64 [[DEF]], [[DEF]], implicit-def $scc
+    ; GFX7-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $scc
+    ; GFX7-NEXT: $sgpr0 = COPY [[COPY]]
+    ; GFX7-NEXT: S_ENDPGM 0, implicit $sgpr0
+    ;
+    ; GF8-LABEL: name: test_copy_scc_vcc
+    ; GF8: [[DEF:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; GF8-NEXT: S_CMP_LG_U64 [[DEF]], 0, implicit-def $scc
+    ; GF8-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $scc
+    ; GF8-NEXT: $sgpr0 = COPY [[COPY]]
+    ; GF8-NEXT: S_ENDPGM 0, implicit $sgpr0
+    ;
+    ; GFX11-LABEL: name: test_copy_scc_vcc
+    ; GFX11: [[DEF:%[0-9]+]]:sreg_32_xm0_xexec = IMPLICIT_DEF
+    ; GFX11-NEXT: S_CMP_LG_U32 [[DEF]], 0, implicit-def $scc
+    ; GFX11-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $scc
+    ; GFX11-NEXT: $sgpr0 = COPY [[COPY]]
+    ; GFX11-NEXT: S_ENDPGM 0, implicit $sgpr0
+    %0:vcc(s1) = G_IMPLICIT_DEF
+    %1:sgpr(s32) = G_AMDGPU_COPY_SCC_VCC %0
+    $sgpr0 = COPY %1
+    S_ENDPGM 0, implicit $sgpr0
+...