[AMDGPU] LCSSA pass added in preISel.

[alex-t]

Summary

All the loops should come into ISel in LCSSA form to mark divergent
uniform values defined in the divergent loop and used outside

[arsenm]

llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp

383

Maybe this should be under EXPENSIVE_CHECKs?

383

The pass dependency in particular

[nhaehnle]

Thanks for dealing with this. Matt's suggestion is reasonable to me, either way LGTM.

[alex-t]

isLCSSAForm check is under EXPENSIVE_CHECKS

llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp

 #include "llvm/CodeGen/ISDOpcodes.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/CodeGen/SelectionDAGISel.h"
 #include "llvm/CodeGen/SelectionDAGNodes.h"
 #include "llvm/CodeGen/ValueTypes.h"
 #include "llvm/IR/BasicBlock.h"
+#ifdef EXPENSIVE_CHECKS
+#include "llvm/IR/Dominators.h"
+#endif
 #include "llvm/IR/Instruction.h"
 #include "llvm/MC/MCInstrDesc.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/CodeGen.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/MachineValueType.h"
 #include "llvm/Support/MathExtras.h"
 #include <cassert>
explicit AMDGPUDAGToDAGISel(TargetMachine *TM = nullptr,
     EnableLateStructurizeCFG = AMDGPUTargetMachine::EnableLateStructurizeCFG;
   }
   ~AMDGPUDAGToDAGISel() override = default;
 
   void getAnalysisUsage(AnalysisUsage &AU) const override {
     AU.addRequired<AMDGPUArgumentUsageInfo>();
     AU.addRequired<AMDGPUPerfHintAnalysis>();
     AU.addRequired<LegacyDivergenceAnalysis>();
+#ifndef EXPENSIVE_CHECKS
+    AU.addRequired<DominatorTreeWrapperPass>();
+    AU.addRequired<LoopInfoWrapperPass>();
+#endif
     SelectionDAGISel::getAnalysisUsage(AU);
   }
 
   bool matchLoadD16FromBuildVector(SDNode *N) const;
 
   bool runOnMachineFunction(MachineFunction &MF) override;
   void PreprocessISelDAG() override;
   void Select(SDNode *N) override;

 
 }  // end anonymous namespace
 
 INITIALIZE_PASS_BEGIN(AMDGPUDAGToDAGISel, "amdgpu-isel",
                       "AMDGPU DAG->DAG Pattern Instruction Selection", false, false)
 INITIALIZE_PASS_DEPENDENCY(AMDGPUArgumentUsageInfo)
 INITIALIZE_PASS_DEPENDENCY(AMDGPUPerfHintAnalysis)
 INITIALIZE_PASS_DEPENDENCY(LegacyDivergenceAnalysis)
+#ifdef EXPENSIVE_CHECKS
+INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
+INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
+#endif
 INITIALIZE_PASS_END(AMDGPUDAGToDAGISel, "amdgpu-isel",
                     "AMDGPU DAG->DAG Pattern Instruction Selection", false, false)
 
 /// This pass converts a legalized DAG into a AMDGPU-specific
 // DAG, ready for instruction scheduling.
 FunctionPass *llvm::createAMDGPUISelDag(TargetMachine *TM,
                                         CodeGenOpt::Level OptLevel) {
   return new AMDGPUDAGToDAGISel(TM, OptLevel);
 }
 
 /// This pass converts a legalized DAG into a R600-specific
 // DAG, ready for instruction scheduling.
 FunctionPass *llvm::createR600ISelDag(TargetMachine *TM,
                                       CodeGenOpt::Level OptLevel) {
   return new R600DAGToDAGISel(TM, OptLevel);
 }
 
 bool AMDGPUDAGToDAGISel::runOnMachineFunction(MachineFunction &MF) {
+#ifdef EXPENSIVE_CHECKS
+  DominatorTree & DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree();
+  LoopInfo * LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
+  for (auto &L : LI->getLoopsInPreorder()) {
+    assert(L->isLCSSAForm(DT));
+  }
+#endif
   Subtarget = &MF.getSubtarget<GCNSubtarget>();
   return SelectionDAGISel::runOnMachineFunction(MF);
 }
 
 bool AMDGPUDAGToDAGISel::matchLoadD16FromBuildVector(SDNode *N) const {
   assert(Subtarget->d16PreservesUnusedBits());
   MVT VT = N->getValueType(0).getSimpleVT();
   if (VT != MVT::v2i16 && VT != MVT::v2f16)

llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp

bool GCNPassConfig::addPreISel() {
   if (!LateCFGStructurize) {
     addPass(createStructurizeCFGPass(true)); // true -> SkipUniformRegions
   }
   addPass(createSinkingPass());
   addPass(createAMDGPUAnnotateUniformValues());
   if (!LateCFGStructurize) {
     addPass(createSIAnnotateControlFlowPass());
   }
+  addPass(createLCSSAPass());
 
   return false;
 }
 
 void GCNPassConfig::addMachineSSAOptimization() {
   TargetPassConfig::addMachineSSAOptimization();
 
   // We want to fold operands after PeepholeOptimizer has run (or as part of

llvm/test/CodeGen/AMDGPU/divergent-branch-uniform-condition.ll

 ; CHECK-LABEL: main:
 ; CHECK:       ; %bb.0: ; %start
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
 ; CHECK-NEXT:    s_mov_b32 m0, s0
 ; CHECK-NEXT:    s_mov_b64 s[4:5], 0
 ; CHECK-NEXT:    v_interp_p1_f32_e32 v0, v1, attr0.x
 ; CHECK-NEXT:    v_cmp_nlt_f32_e64 s[0:1], 0, v0
 ; CHECK-NEXT:    v_mov_b32_e32 v1, 0
-; CHECK-NEXT:    ; implicit-def: $sgpr2_sgpr3
+; CHECK-NEXT:    ; implicit-def: $sgpr8_sgpr9
 ; CHECK-NEXT:    ; implicit-def: $sgpr6_sgpr7
+; CHECK-NEXT:    ; implicit-def: $sgpr2_sgpr3
 ; CHECK-NEXT:    s_branch BB0_3
-; CHECK-NEXT:  BB0_1: ; in Loop: Header=BB0_3 Depth=1
-; CHECK-NEXT:    ; implicit-def: $vgpr1
+; CHECK-NEXT:  BB0_1: ; %Flow1
+; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
+; CHECK-NEXT:    s_or_b64 exec, exec, s[8:9]
+; CHECK-NEXT:    s_mov_b64 s[8:9], 0
 ; CHECK-NEXT:  BB0_2: ; %Flow
 ; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
-; CHECK-NEXT:    s_and_b64 s[8:9], exec, s[6:7]
-; CHECK-NEXT:    s_or_b64 s[8:9], s[8:9], s[4:5]
-; CHECK-NEXT:    s_mov_b64 s[4:5], s[8:9]
-; CHECK-NEXT:    s_andn2_b64 exec, exec, s[8:9]
-; CHECK-NEXT:    s_cbranch_execz BB0_7
+; CHECK-NEXT:    s_and_b64 s[10:11], exec, s[6:7]
+; CHECK-NEXT:    s_or_b64 s[10:11], s[10:11], s[4:5]
+; CHECK-NEXT:    s_andn2_b64 s[2:3], s[2:3], exec
+; CHECK-NEXT:    s_and_b64 s[4:5], s[8:9], exec
+; CHECK-NEXT:    s_or_b64 s[2:3], s[2:3], s[4:5]
+; CHECK-NEXT:    s_mov_b64 s[4:5], s[10:11]
+; CHECK-NEXT:    s_andn2_b64 exec, exec, s[10:11]
+; CHECK-NEXT:    s_cbranch_execz BB0_6
 ; CHECK-NEXT:  BB0_3: ; %loop
 ; CHECK-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v1
 ; CHECK-NEXT:    s_and_b64 vcc, exec, vcc
 ; CHECK-NEXT:    s_or_b64 s[6:7], s[6:7], exec
-; CHECK-NEXT:    s_or_b64 s[2:3], s[2:3], exec
-; CHECK-NEXT:    s_cbranch_vccz BB0_1
+; CHECK-NEXT:    s_or_b64 s[8:9], s[8:9], exec
+; CHECK-NEXT:    s_cbranch_vccz BB0_2
 ; CHECK-NEXT:  ; %bb.4: ; %endif1
 ; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
 ; CHECK-NEXT:    s_mov_b64 s[6:7], -1
 ; CHECK-NEXT:    s_and_saveexec_b64 s[8:9], s[0:1]
 ; CHECK-NEXT:    s_xor_b64 s[8:9], exec, s[8:9]
-; CHECK-NEXT:    ; mask branch BB0_6
+; CHECK-NEXT:    ; mask branch BB0_1
+; CHECK-NEXT:    s_cbranch_execz BB0_1
 ; CHECK-NEXT:  BB0_5: ; %endif2
 ; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
 ; CHECK-NEXT:    v_add_u32_e32 v1, 1, v1
 ; CHECK-NEXT:    s_xor_b64 s[6:7], exec, -1
-; CHECK-NEXT:  BB0_6: ; %Flow1
-; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
-; CHECK-NEXT:    s_or_b64 exec, exec, s[8:9]
-; CHECK-NEXT:    s_andn2_b64 s[2:3], s[2:3], exec
-; CHECK-NEXT:    s_branch BB0_2
-; CHECK-NEXT:  BB0_7: ; %Flow2
-; CHECK-NEXT:    s_or_b64 exec, exec, s[8:9]
+; CHECK-NEXT:    s_branch BB0_1
+; CHECK-NEXT:  BB0_6: ; %Flow2
+; CHECK-NEXT:    s_or_b64 exec, exec, s[10:11]
 ; CHECK-NEXT:    v_mov_b32_e32 v1, 0
-; this is the divergent branch with the condition not marked as divergent
 ; CHECK-NEXT:    s_and_saveexec_b64 s[0:1], s[2:3]
-; CHECK-NEXT:    ; mask branch BB0_9
-; CHECK-NEXT:  BB0_8: ; %if1
+; CHECK-NEXT:    ; mask branch BB0_8
+; CHECK-NEXT:  BB0_7: ; %if1
 ; CHECK-NEXT:    v_sqrt_f32_e32 v1, v0
-; CHECK-NEXT:  BB0_9: ; %endloop
+; CHECK-NEXT:  BB0_8: ; %endloop
 ; CHECK-NEXT:    s_or_b64 exec, exec, s[0:1]
 ; CHECK-NEXT:    exp mrt0 v1, v1, v1, v1 done vm
 ; CHECK-NEXT:    s_endpgm
+; this is the divergent branch with the condition not marked as divergent
 start:
   %v0 = call float @llvm.amdgcn.interp.p1(float %1, i32 0, i32 0, i32 %0)
   br label %loop
 
 loop:
   %v1 = phi i32 [ 0, %start ], [ %v5, %endif2 ]
   %v2 = icmp ugt i32 %v1, 31
   br i1 %v2, label %if1, label %endif1