Index: lib/Transforms/Scalar/LoopPredication.cpp =================================================================== --- lib/Transforms/Scalar/LoopPredication.cpp +++ lib/Transforms/Scalar/LoopPredication.cpp @@ -228,6 +228,14 @@ "expressed as widenable branches to deoptimize blocks"), cl::init(true)); +// EXPERIMENTAL - See comment below about how this is currently UNSOUND as +// currently implemented. +static cl::opt PredicateLoopExits( + "loop-predication-all-branches", cl::Hidden, + cl::desc("Attempt to predicate conditions in normal branches"), + cl::init(false)); + + namespace { class LoopPredication { /// Represents an induction variable check: @@ -282,7 +290,7 @@ unsigned collectChecks(SmallVectorImpl &Checks, Value *Condition, SCEVExpander &Expander, IRBuilder<> &Builder); bool widenGuardConditions(IntrinsicInst *II, SCEVExpander &Expander); - bool widenWidenableBranchGuardConditions(BranchInst *Guard, SCEVExpander &Expander); + bool widenBranchConditions(BranchInst *BI, SCEVExpander &Expander); // If the loop always exits through another block in the loop, we should not // predicate based on the latch check. For example, the latch check can be a // very coarse grained check and there can be more fine grained exit checks @@ -667,10 +675,12 @@ return true; } -bool LoopPredication::widenWidenableBranchGuardConditions( - BranchInst *BI, SCEVExpander &Expander) { - assert(isGuardAsWidenableBranch(BI) && "Must be!"); - LLVM_DEBUG(dbgs() << "Processing guard:\n"); +bool LoopPredication::widenBranchConditions(BranchInst *BI, + SCEVExpander &Expander) { +#ifndef NDEBUG + const bool WasWidenableBranch = isGuardAsWidenableBranch(BI); +#endif + LLVM_DEBUG(dbgs() << "Processing branch:\n"); LLVM_DEBUG(BI->dump()); TotalConsidered++; @@ -693,7 +703,7 @@ LastCheck = Builder.CreateAnd(LastCheck, Check); auto *OldCond = BI->getCondition(); BI->setCondition(LastCheck); - assert(isGuardAsWidenableBranch(BI) && + assert((!WasWidenableBranch || isGuardAsWidenableBranch(BI)) && "Stopped being a guard after transform?"); RecursivelyDeleteTriviallyDeadInstructions(OldCond); @@ -863,7 +873,7 @@ Intrinsic::getName(Intrinsic::experimental_widenable_condition)); bool HasWidenableConditions = PredicateWidenableBranchGuards && WCDecl && !WCDecl->use_empty(); - if (!HasIntrinsicGuards && !HasWidenableConditions) + if (!HasIntrinsicGuards && !HasWidenableConditions && !PredicateLoopExits) return false; DL = &M->getDataLayout(); @@ -884,21 +894,58 @@ LLVM_DEBUG(dbgs() << "Loop not profitable to predicate!\n"); return false; } + + Optional HasSideEffectsCache = None; + auto HasSideEffects = [&]() { + if (!HasSideEffectsCache) { + bool Res = false; + for (const auto BB : L->blocks()) + for (auto &I : *BB) + Res = Res || I.mayHaveSideEffects(); + HasSideEffectsCache = Res; + } + return HasSideEffectsCache.getValue(); + }; + auto IsExittingBranchWithoutLoopUses = [&](Instruction *TI) { + auto *BI = dyn_cast(TI); + if (!BI || BI->isUnconditional()) + return false; + // Expect second to be outside loop (TODO: generalize) + if (!L->contains(BI->getSuccessor(0)) || L->contains(BI->getSuccessor(1))) + return false; + + // TODO: must control backedge! (part of neccessary condition to ensure no + // false triggers) + + // LCSSA form + No PHIs -> no uses of values computed in loop + //assert(L->isLCSSAForm(*DT) && "Loop is not in LCSSA form."); + auto phis = BI->getSuccessor(1)->phis(); + return phis.begin() == phis.end(); + }; + // Collect all the guards into a vector and process later, so as not // to invalidate the instruction iterator. SmallVector Guards; - SmallVector GuardsAsWidenableBranches; + SmallVector Branches;; for (const auto BB : L->blocks()) { for (auto &I : *BB) if (isGuard(&I)) Guards.push_back(cast(&I)); if (PredicateWidenableBranchGuards && isGuardAsWidenableBranch(BB->getTerminator())) - GuardsAsWidenableBranches.push_back( - cast(BB->getTerminator())); + Branches.push_back(cast(BB->getTerminator())); + else if (PredicateLoopExits && !HasSideEffects() && + IsExittingBranchWithoutLoopUses(BB->getTerminator())) + // Note: This bit of code is currently expiremental, and is known to be + // incorrect in one important way. While changing which iteration we + // exit at is legal for a normal branch, changing whether we exit or not + // would be incorrect. The code for guards gets to spuriously cause a + // condition to fail; the branch version does not. None of the code has + // been auditted (or tested) for the stricter invariant just yet. + Branches.push_back(cast(BB->getTerminator())); } - if (Guards.empty() && GuardsAsWidenableBranches.empty()) + if (Guards.empty() && Branches.empty()) return false; SCEVExpander Expander(*SE, *DL, "loop-predication"); @@ -906,8 +953,8 @@ bool Changed = false; for (auto *Guard : Guards) Changed |= widenGuardConditions(Guard, Expander); - for (auto *Guard : GuardsAsWidenableBranches) - Changed |= widenWidenableBranchGuardConditions(Guard, Expander); + for (auto *BI : Branches) + Changed |= widenBranchConditions(BI, Expander); return Changed; } Index: test/Transforms/LoopPredication/standard_branches.ll =================================================================== --- test/Transforms/LoopPredication/standard_branches.ll +++ test/Transforms/LoopPredication/standard_branches.ll @@ -0,0 +1,1757 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py +; RUN: opt -S -loop-predication -loop-predication-all-branches=true < %s 2>&1 | FileCheck %s +; RUN: opt -S -passes='require,loop(loop-predication)' -loop-predication-all-branches=true < %s 2>&1 | FileCheck %s + +define i32 @unsigned_loop_0_to_n_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @unsigned_loop_0_to_n_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @unsigned_loop_0_to_n_ule_latch_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @unsigned_loop_0_to_n_ule_latch_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ule i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ule i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @unsigned_loop_0_to_n_ugt_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @unsigned_loop_0_to_n_ugt_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ugt i32 %length, %i + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp sle i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; + +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_ult_check_length_range_known(i32* %array, i32* %length.ptr, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_ult_check_length_range_known( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: [[LENGTH:%.*]] = load i32, i32* [[LENGTH_PTR:%.*]], !range !1 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp sle i32 [[N]], [[LENGTH]] +; CHECK-NEXT: [[TMP1:%.*]] = and i1 true, [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP1]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + %length = load i32, i32* %length.ptr, !range !1 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_inverse_latch_predicate(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_inverse_latch_predicate( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp slt i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp sgt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[EXIT_LOOPEXIT:%.*]], label [[LOOP]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp sgt i32 %i.next, %n + br i1 %continue, label %exit, label %loop + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_sle_latch_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_sle_latch_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp slt i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp sle i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp sle i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_preincrement_latch_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_preincrement_latch_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[LENGTH:%.*]], -1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp sle i32 [[N]], [[TMP0]] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP2]], [[TMP1]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP3]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add i32 %i, 1 + %continue = icmp slt i32 %i, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_preincrement_latch_check_postincrement_guard_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_preincrement_latch_check_postincrement_guard_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[LENGTH:%.*]], -2 +; CHECK-NEXT: [[TMP1:%.*]] = icmp sle i32 [[N]], [[TMP0]] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 1, [[LENGTH]] +; CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP2]], [[TMP1]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: br i1 [[TMP3]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %i.next = add i32 %i, 1 + %within.bounds = icmp ult i32 %i.next, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %continue = icmp slt i32 %i, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_sle_latch_offset_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_sle_latch_offset_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[LENGTH:%.*]], -1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp slt i32 [[N]], [[TMP0]] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 1, [[LENGTH]] +; CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP2]], [[TMP1]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP3]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp sle i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %i.offset = add i32 %i, 1 + %within.bounds = icmp ult i32 %i.offset, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add i32 %i, 1 + %continue = icmp sle i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_offset_sle_latch_offset_ult_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_offset_sle_latch_offset_ult_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp slt i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 1, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: [[I_NEXT_OFFSET:%.*]] = add i32 [[I_NEXT]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp sle i32 [[I_NEXT_OFFSET]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %i.offset = add i32 %i, 1 + %within.bounds = icmp ult i32 %i.offset, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add i32 %i, 1 + %i.next.offset = add i32 %i.next, 1 + %continue = icmp sle i32 %i.next.offset, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @unsupported_latch_pred_loop_0_to_n(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @unsupported_latch_pred_loop_0_to_n( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i32 [[I]], [[LENGTH:%.*]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ne i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nsw i32 %i, 1 + %continue = icmp ne i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_unsupported_iv_step(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_unsupported_iv_step( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i32 [[I]], [[LENGTH:%.*]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 2 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nsw i32 %i, 2 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_equal_iv_range_check(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_equal_iv_range_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp sle i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[J:%.*]] = phi i32 [ [[J_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[J_NEXT]] = add nsw i32 [[J]], 1 +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %j = phi i32 [ %j.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %j, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %j.next = add nsw i32 %j, 1 + %i.next = add nsw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_start_to_n_offset_iv_range_check(i32* %array, i32 %start.i, i32 %start.j, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_start_to_n_offset_iv_range_check( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[LENGTH:%.*]], [[START_I:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = sub i32 [[TMP0]], [[START_J:%.*]] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sle i32 [[N]], [[TMP1]] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ult i32 [[START_J]], [[LENGTH]] +; CHECK-NEXT: [[TMP4:%.*]] = and i1 [[TMP3]], [[TMP2]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ [[START_I]], [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[J:%.*]] = phi i32 [ [[J_NEXT:%.*]], [[GUARDED]] ], [ [[START_J]], [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP4]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[J_NEXT]] = add i32 [[J]], 1 +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ %start.i, %loop.preheader ] + %j = phi i32 [ %j.next, %guarded ], [ %start.j, %loop.preheader ] + %within.bounds = icmp ult i32 %j, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %j.next = add i32 %j, 1 + %i.next = add i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_different_iv_types(i32* %array, i16 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_different_iv_types( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[J:%.*]] = phi i16 [ [[J_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i16 [[J]], [[LENGTH:%.*]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[J_NEXT]] = add i16 [[J]], 1 +; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %j = phi i16 [ %j.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i16 %j, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %j.next = add i16 %j, 1 + %i.next = add i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_0_to_n_different_iv_strides(i32* %array, i32 %length, i32 %n) { +; CHECK-LABEL: @signed_loop_0_to_n_different_iv_strides( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[J:%.*]] = phi i32 [ [[J_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i32 [[J]], [[LENGTH:%.*]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[J_NEXT]] = add nsw i32 [[J]], 2 +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %j = phi i32 [ %j.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %j, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %j.next = add nsw i32 %j, 2 + %i.next = add nsw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @two_range_checks(i32* %array.1, i32 %length.1, i32* %array.2, i32 %length.2, i32 %n) { +; CHECK-LABEL: @two_range_checks( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH_2:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH_2]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ule i32 [[N]], [[LENGTH_1:%.*]] +; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 0, [[LENGTH_1]] +; CHECK-NEXT: [[TMP5:%.*]] = and i1 [[TMP4]], [[TMP3]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[TMP6:%.*]] = and i1 [[TMP2]], [[TMP5]] +; CHECK-NEXT: br i1 [[TMP6]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_1_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_1:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_1_I:%.*]] = load i32, i32* [[ARRAY_1_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_1:%.*]] = add i32 [[LOOP_ACC]], [[ARRAY_1_I]] +; CHECK-NEXT: [[ARRAY_2_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_2:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_2_I:%.*]] = load i32, i32* [[ARRAY_2_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC_1]], [[ARRAY_2_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds.1 = icmp ult i32 %i, %length.1 + %within.bounds.2 = icmp ult i32 %i, %length.2 + %within.bounds = and i1 %within.bounds.1, %within.bounds.2 + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.1.i.ptr = getelementptr inbounds i32, i32* %array.1, i64 %i.i64 + %array.1.i = load i32, i32* %array.1.i.ptr, align 4 + %loop.acc.1 = add i32 %loop.acc, %array.1.i + %array.2.i.ptr = getelementptr inbounds i32, i32* %array.2, i64 %i.i64 + %array.2.i = load i32, i32* %array.2.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc.1, %array.2.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @three_range_checks(i32* %array.1, i32 %length.1, i32* %array.2, i32 %length.2, i32* %array.3, i32 %length.3, i32 %n) { +; CHECK-LABEL: @three_range_checks( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH_3:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH_3]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ule i32 [[N]], [[LENGTH_2:%.*]] +; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 0, [[LENGTH_2]] +; CHECK-NEXT: [[TMP5:%.*]] = and i1 [[TMP4]], [[TMP3]] +; CHECK-NEXT: [[TMP6:%.*]] = icmp ule i32 [[N]], [[LENGTH_1:%.*]] +; CHECK-NEXT: [[TMP7:%.*]] = icmp ult i32 0, [[LENGTH_1]] +; CHECK-NEXT: [[TMP8:%.*]] = and i1 [[TMP7]], [[TMP6]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[TMP9:%.*]] = and i1 [[TMP2]], [[TMP5]] +; CHECK-NEXT: [[TMP10:%.*]] = and i1 [[TMP9]], [[TMP8]] +; CHECK-NEXT: br i1 [[TMP10]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_1_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_1:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_1_I:%.*]] = load i32, i32* [[ARRAY_1_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_1:%.*]] = add i32 [[LOOP_ACC]], [[ARRAY_1_I]] +; CHECK-NEXT: [[ARRAY_2_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_2:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_2_I:%.*]] = load i32, i32* [[ARRAY_2_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_2:%.*]] = add i32 [[LOOP_ACC_1]], [[ARRAY_2_I]] +; CHECK-NEXT: [[ARRAY_3_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_3:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_3_I:%.*]] = load i32, i32* [[ARRAY_3_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC_2]], [[ARRAY_3_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds.1 = icmp ult i32 %i, %length.1 + %within.bounds.2 = icmp ult i32 %i, %length.2 + %within.bounds.3 = icmp ult i32 %i, %length.3 + %within.bounds.1.and.2 = and i1 %within.bounds.1, %within.bounds.2 + %within.bounds = and i1 %within.bounds.1.and.2, %within.bounds.3 + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.1.i.ptr = getelementptr inbounds i32, i32* %array.1, i64 %i.i64 + %array.1.i = load i32, i32* %array.1.i.ptr, align 4 + %loop.acc.1 = add i32 %loop.acc, %array.1.i + %array.2.i.ptr = getelementptr inbounds i32, i32* %array.2, i64 %i.i64 + %array.2.i = load i32, i32* %array.2.i.ptr, align 4 + %loop.acc.2 = add i32 %loop.acc.1, %array.2.i + %array.3.i.ptr = getelementptr inbounds i32, i32* %array.3, i64 %i.i64 + %array.3.i = load i32, i32* %array.3.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc.2, %array.3.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @three_guards(i32* %array.1, i32 %length.1, i32* %array.2, i32 %length.2, i32* %array.3, i32 %length.3, i32 %n) { +; CHECK-LABEL: @three_guards( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH_1:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH_1]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ule i32 [[N]], [[LENGTH_2:%.*]] +; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 0, [[LENGTH_2]] +; CHECK-NEXT: [[TMP5:%.*]] = and i1 [[TMP4]], [[TMP3]] +; CHECK-NEXT: [[TMP6:%.*]] = icmp ule i32 [[N]], [[LENGTH_3:%.*]] +; CHECK-NEXT: [[TMP7:%.*]] = icmp ult i32 0, [[LENGTH_3]] +; CHECK-NEXT: [[TMP8:%.*]] = and i1 [[TMP7]], [[TMP6]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED6:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED6]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP2]], label [[GUARDED:%.*]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_1_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_1:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_1_I:%.*]] = load i32, i32* [[ARRAY_1_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_1:%.*]] = add i32 [[LOOP_ACC]], [[ARRAY_1_I]] +; CHECK-NEXT: br i1 [[TMP5]], label [[GUARDED1:%.*]], label [[DEOPT2:%.*]], !prof !0 +; CHECK: deopt2: +; CHECK-NEXT: [[DEOPTCALL3:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL3]] +; CHECK: guarded1: +; CHECK-NEXT: [[ARRAY_2_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_2:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_2_I:%.*]] = load i32, i32* [[ARRAY_2_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_2:%.*]] = add i32 [[LOOP_ACC_1]], [[ARRAY_2_I]] +; CHECK-NEXT: br i1 [[TMP8]], label [[GUARDED6]], label [[DEOPT7:%.*]], !prof !0 +; CHECK: deopt7: +; CHECK-NEXT: [[DEOPTCALL8:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL8]] +; CHECK: guarded6: +; CHECK-NEXT: [[ARRAY_3_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY_3:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_3_I:%.*]] = load i32, i32* [[ARRAY_3_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC_2]], [[ARRAY_3_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED6]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded6, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded6 ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded6 ], [ 0, %loop.preheader ] + %within.bounds.1 = icmp ult i32 %i, %length.1 + br i1 %within.bounds.1, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.1.i.ptr = getelementptr inbounds i32, i32* %array.1, i64 %i.i64 + %array.1.i = load i32, i32* %array.1.i.ptr, align 4 + %loop.acc.1 = add i32 %loop.acc, %array.1.i + %within.bounds.2 = icmp ult i32 %i, %length.2 + br i1 %within.bounds.2, label %guarded1, label %deopt2, !prof !0 + +deopt2: ; preds = %guarded + %deoptcall3 = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall3 + +guarded1: ; preds = %guarded + %array.2.i.ptr = getelementptr inbounds i32, i32* %array.2, i64 %i.i64 + %array.2.i = load i32, i32* %array.2.i.ptr, align 4 + %loop.acc.2 = add i32 %loop.acc.1, %array.2.i + %within.bounds.3 = icmp ult i32 %i, %length.3 + br i1 %within.bounds.3, label %guarded6, label %deopt7, !prof !0 + +deopt7: ; preds = %guarded1 + %deoptcall8 = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall8 + +guarded6: ; preds = %guarded1 + %array.3.i.ptr = getelementptr inbounds i32, i32* %array.3, i64 %i.i64 + %array.3.i = load i32, i32* %array.3.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc.2, %array.3.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded6, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded6 ] + ret i32 %result +} + +define i32 @unsigned_loop_0_to_n_unrelated_condition(i32* %array, i32 %length, i32 %n, i32 %x) { +; CHECK-LABEL: @unsigned_loop_0_to_n_unrelated_condition( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = icmp ule i32 [[N]], [[LENGTH:%.*]] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 0, [[LENGTH]] +; CHECK-NEXT: [[TMP2:%.*]] = and i1 [[TMP1]], [[TMP0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[UNRELATED_COND:%.*]] = icmp ult i32 [[X:%.*]], [[LENGTH]] +; CHECK-NEXT: [[TMP3:%.*]] = and i1 [[UNRELATED_COND]], [[TMP2]] +; CHECK-NEXT: br i1 [[TMP3]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %within.bounds = icmp ult i32 %i, %length + %unrelated.cond = icmp ult i32 %x, %length + %guard.cond = and i1 %within.bounds, %unrelated.cond + br i1 %guard.cond, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @test_no_widened_conditions(i32* %array, i32 %length, i32 %n, i32 %x1, i32 %x2, i32 %x3) { +; CHECK-LABEL: @test_no_widened_conditions( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[UNRELATED_COND_1:%.*]] = icmp eq i32 [[X1:%.*]], [[I]] +; CHECK-NEXT: [[UNRELATED_COND_2:%.*]] = icmp eq i32 [[X2:%.*]], [[I]] +; CHECK-NEXT: [[UNRELATED_COND_3:%.*]] = icmp eq i32 [[X3:%.*]], [[I]] +; CHECK-NEXT: [[UNRELATED_COND_AND_1:%.*]] = and i1 [[UNRELATED_COND_1]], [[UNRELATED_COND_2]] +; CHECK-NEXT: [[GUARD_COND:%.*]] = and i1 [[UNRELATED_COND_AND_1]], [[UNRELATED_COND_3]] +; CHECK-NEXT: br i1 [[GUARD_COND]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %unrelated.cond.1 = icmp eq i32 %x1, %i + %unrelated.cond.2 = icmp eq i32 %x2, %i + %unrelated.cond.3 = icmp eq i32 %x3, %i + %unrelated.cond.and.1 = and i1 %unrelated.cond.1, %unrelated.cond.2 + %guard.cond = and i1 %unrelated.cond.and.1, %unrelated.cond.3 + br i1 %guard.cond, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_start_to_n_loop_variant_bound(i32* %array, i32 %x, i32 %start, i32 %n) { +; CHECK-LABEL: @signed_loop_start_to_n_loop_variant_bound( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ [[START:%.*]], [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[BOUND:%.*]] = add i32 [[I]], [[X:%.*]] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i32 [[I]], [[BOUND]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ %start, %loop.preheader ] + %bound = add i32 %i, %x + %within.bounds = icmp ult i32 %i, %bound + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nsw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @signed_loop_start_to_n_non_monotonic_predicate(i32* %array, i32 %x, i32 %start, i32 %n) { +; CHECK-LABEL: @signed_loop_start_to_n_non_monotonic_predicate( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp sle i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ [[START:%.*]], [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[GUARD_COND:%.*]] = icmp eq i32 [[I]], [[X:%.*]] +; CHECK-NEXT: br i1 [[GUARD_COND]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp slt i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp sle i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ %start, %loop.preheader ] + %guard.cond = icmp eq i32 %i, %x + br i1 %guard.cond, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nsw i32 %i, 1 + %continue = icmp slt i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @unsigned_loop_0_to_n_hoist_length(i32* %array, i16 %length.i16, i32 %n) { +; CHECK-LABEL: @unsigned_loop_0_to_n_hoist_length( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = zext i16 [[LENGTH_I16:%.*]] to i32 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ule i32 [[N]], [[TMP0]] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 0, [[TMP0]] +; CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP2]], [[TMP1]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: br i1 [[TMP3]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %length = zext i16 %length.i16 to i32 + %within.bounds = icmp ult i32 %i, %length + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +define i32 @unsigned_loop_0_to_n_cant_hoist_length(i32* %array, i32 %length, i32 %divider, i32 %n) { +; CHECK-LABEL: @unsigned_loop_0_to_n_cant_hoist_length( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[N:%.*]], 0 +; CHECK-NEXT: br i1 [[TMP5]], label [[EXIT:%.*]], label [[LOOP_PREHEADER:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[LOOP_ACC:%.*]] = phi i32 [ [[LOOP_ACC_NEXT:%.*]], [[GUARDED:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[I_NEXT:%.*]], [[GUARDED]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[LENGTH_UDIV:%.*]] = udiv i32 [[LENGTH:%.*]], [[DIVIDER:%.*]] +; CHECK-NEXT: [[WITHIN_BOUNDS:%.*]] = icmp ult i32 [[I]], [[LENGTH_UDIV]] +; CHECK-NEXT: br i1 [[WITHIN_BOUNDS]], label [[GUARDED]], label [[DEOPT:%.*]], !prof !0 +; CHECK: deopt: +; CHECK-NEXT: [[DEOPTCALL:%.*]] = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] +; CHECK-NEXT: ret i32 [[DEOPTCALL]] +; CHECK: guarded: +; CHECK-NEXT: [[I_I64:%.*]] = zext i32 [[I]] to i64 +; CHECK-NEXT: [[ARRAY_I_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARRAY:%.*]], i64 [[I_I64]] +; CHECK-NEXT: [[ARRAY_I:%.*]] = load i32, i32* [[ARRAY_I_PTR]], align 4 +; CHECK-NEXT: [[LOOP_ACC_NEXT]] = add i32 [[LOOP_ACC]], [[ARRAY_I]] +; CHECK-NEXT: [[I_NEXT]] = add nuw i32 [[I]], 1 +; CHECK-NEXT: [[CONTINUE:%.*]] = icmp ult i32 [[I_NEXT]], [[N]] +; CHECK-NEXT: br i1 [[CONTINUE]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: [[LOOP_ACC_NEXT_LCSSA:%.*]] = phi i32 [ [[LOOP_ACC_NEXT]], [[GUARDED]] ] +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: [[RESULT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[LOOP_ACC_NEXT_LCSSA]], [[EXIT_LOOPEXIT]] ] +; CHECK-NEXT: ret i32 [[RESULT]] +; +entry: + %tmp5 = icmp eq i32 %n, 0 + br i1 %tmp5, label %exit, label %loop.preheader + +loop.preheader: ; preds = %entry + br label %loop + +loop: ; preds = %guarded, %loop.preheader + %loop.acc = phi i32 [ %loop.acc.next, %guarded ], [ 0, %loop.preheader ] + %i = phi i32 [ %i.next, %guarded ], [ 0, %loop.preheader ] + %length.udiv = udiv i32 %length, %divider + %within.bounds = icmp ult i32 %i, %length.udiv + br i1 %within.bounds, label %guarded, label %deopt, !prof !0 + +deopt: ; preds = %loop + %deoptcall = call i32 (...) @llvm.experimental.deoptimize.i32(i32 9) [ "deopt"() ] + ret i32 %deoptcall + +guarded: ; preds = %loop + %i.i64 = zext i32 %i to i64 + %array.i.ptr = getelementptr inbounds i32, i32* %array, i64 %i.i64 + %array.i = load i32, i32* %array.i.ptr, align 4 + %loop.acc.next = add i32 %loop.acc, %array.i + %i.next = add nuw i32 %i, 1 + %continue = icmp ult i32 %i.next, %n + br i1 %continue, label %loop, label %exit + +exit: ; preds = %guarded, %entry + %result = phi i32 [ 0, %entry ], [ %loop.acc.next, %guarded ] + ret i32 %result +} + +declare i32 @llvm.experimental.deoptimize.i32(...) + +attributes #0 = { inaccessiblememonly nounwind } + +!0 = !{!"branch_weights", i32 1048576, i32 1} +!1 = !{i32 1, i32 -2147483648}