Allow to generate a loop without the GuardBB
This allows us to omit the GuardBB in front of created loops
if we can show the loop trip count is at least one. It also
simplifies the dominance relation inside the new created region.
A GuardBB (even with a constant branch condition) might trigger
false dominance errors during function verification.
Differential Revision: http://reviews.llvm.org/D5297
llvm-svn: 217525
diff --git a/polly/include/polly/CodeGen/LoopGenerators.h b/polly/include/polly/CodeGen/LoopGenerators.h
index 6467161..6f766e4 100644
--- a/polly/include/polly/CodeGen/LoopGenerators.h
+++ b/polly/include/polly/CodeGen/LoopGenerators.h
@@ -27,7 +27,7 @@
namespace polly {
using namespace llvm;
-/// @brief Create a scalar loop.
+/// @brief Create a scalar do/for-style loop.
///
/// @param LowerBound The starting value of the induction variable.
/// @param UpperBound The upper bound of the induction variable.
@@ -43,12 +43,16 @@
/// @param Annotator This function can (optionally) take a LoopAnnotator which
/// tracks the loop structure.
/// @param Parallel If this loop should be marked parallel in the Annotator.
+/// @param UseGuard Create a guard in front of the header to check if the
+/// loop is executed at least once, otherwise just assume it.
+///
/// @return Value* The newly created induction variable for this loop.
Value *createLoop(Value *LowerBound, Value *UpperBound, Value *Stride,
PollyIRBuilder &Builder, Pass *P, LoopInfo &LI,
DominatorTree &DT, BasicBlock *&ExitBlock,
ICmpInst::Predicate Predicate,
- LoopAnnotator *Annotator = NULL, bool Parallel = false);
+ LoopAnnotator *Annotator = NULL, bool Parallel = false,
+ bool UseGuard = true);
class OMPGenerator {
public:
diff --git a/polly/lib/CodeGen/IslCodeGeneration.cpp b/polly/lib/CodeGen/IslCodeGeneration.cpp
index 84b91ed..0d9a798 100644
--- a/polly/lib/CodeGen/IslCodeGeneration.cpp
+++ b/polly/lib/CodeGen/IslCodeGeneration.cpp
@@ -352,8 +352,12 @@
if (MaxType != ValueInc->getType())
ValueInc = Builder.CreateSExt(ValueInc, MaxType);
+ // If we can show that LB <Predicate> UB holds at least once, we can
+ // omit the GuardBB in front of the loop.
+ bool UseGuardBB =
+ !SE.isKnownPredicate(Predicate, SE.getSCEV(ValueLB), SE.getSCEV(ValueUB));
IV = createLoop(ValueLB, ValueUB, ValueInc, Builder, P, LI, DT, ExitBlock,
- Predicate, &Annotator, Parallel);
+ Predicate, &Annotator, Parallel, UseGuardBB);
IDToValue[IteratorID] = IV;
create(Body);
diff --git a/polly/lib/CodeGen/LoopGenerators.cpp b/polly/lib/CodeGen/LoopGenerators.cpp
index 4eaaac0..267caeb 100644
--- a/polly/lib/CodeGen/LoopGenerators.cpp
+++ b/polly/lib/CodeGen/LoopGenerators.cpp
@@ -23,34 +23,33 @@
using namespace llvm;
using namespace polly;
-// We generate a loop of the following structure
+// We generate a loop of either of the following structures:
//
-// BeforeBB
-// |
-// v
-// GuardBB
-// / |
-// __ PreHeaderBB |
-// / \ / |
-// latch HeaderBB |
-// \ / \ /
-// < \ /
-// \ /
-// ExitBB
+// BeforeBB BeforeBB
+// | |
+// v v
+// GuardBB PreHeaderBB
+// / | | _____
+// __ PreHeaderBB | v \/ |
+// / \ / | HeaderBB latch
+// latch HeaderBB | |\ |
+// \ / \ / | \------/
+// < \ / |
+// \ / v
+// ExitBB ExitBB
//
-// GuardBB checks if the loop is executed at least once. If this is the case
-// we branch to PreHeaderBB and subsequently to the HeaderBB, which contains the
-// loop iv 'polly.indvar', the incremented loop iv 'polly.indvar_next' as well
-// as the condition to check if we execute another iteration of the loop. After
-// the loop has finished, we branch to ExitBB.
-//
-// TODO: We currently always create the GuardBB. If we can prove the loop is
-// always executed at least once, we can get rid of this branch.
+// depending on whether or not we know that it is executed at least once. If
+// not, GuardBB checks if the loop is executed at least once. If this is the
+// case we branch to PreHeaderBB and subsequently to the HeaderBB, which
+// contains the loop iv 'polly.indvar', the incremented loop iv
+// 'polly.indvar_next' as well as the condition to check if we execute another
+// iteration of the loop. After the loop has finished, we branch to ExitBB.
Value *polly::createLoop(Value *LB, Value *UB, Value *Stride,
PollyIRBuilder &Builder, Pass *P, LoopInfo &LI,
DominatorTree &DT, BasicBlock *&ExitBB,
ICmpInst::Predicate Predicate,
- LoopAnnotator *Annotator, bool Parallel) {
+ LoopAnnotator *Annotator, bool Parallel,
+ bool UseGuard) {
Function *F = Builder.GetInsertBlock()->getParent();
LLVMContext &Context = F->getContext();
@@ -59,7 +58,8 @@
assert(LoopIVType && "UB is not integer?");
BasicBlock *BeforeBB = Builder.GetInsertBlock();
- BasicBlock *GuardBB = BasicBlock::Create(Context, "polly.loop_if", F);
+ BasicBlock *GuardBB =
+ UseGuard ? BasicBlock::Create(Context, "polly.loop_if", F) : nullptr;
BasicBlock *HeaderBB = BasicBlock::Create(Context, "polly.loop_header", F);
BasicBlock *PreHeaderBB =
BasicBlock::Create(Context, "polly.loop_preheader", F);
@@ -74,16 +74,15 @@
Loop *OuterLoop = LI.getLoopFor(BeforeBB);
Loop *NewLoop = new Loop();
- if (OuterLoop) {
+ if (OuterLoop)
OuterLoop->addChildLoop(NewLoop);
- } else {
+ else
LI.addTopLevelLoop(NewLoop);
- }
- if (OuterLoop) {
+ if (OuterLoop && GuardBB)
OuterLoop->addBasicBlockToLoop(GuardBB, LI.getBase());
+ else if (OuterLoop)
OuterLoop->addBasicBlockToLoop(PreHeaderBB, LI.getBase());
- }
NewLoop->addBasicBlockToLoop(HeaderBB, LI.getBase());
@@ -92,18 +91,23 @@
ExitBB->setName("polly.loop_exit");
// BeforeBB
- BeforeBB->getTerminator()->setSuccessor(0, GuardBB);
+ if (GuardBB) {
+ BeforeBB->getTerminator()->setSuccessor(0, GuardBB);
+ DT.addNewBlock(GuardBB, BeforeBB);
- // GuardBB
- DT.addNewBlock(GuardBB, BeforeBB);
- Builder.SetInsertPoint(GuardBB);
- Value *LoopGuard;
- LoopGuard = Builder.CreateICmp(Predicate, LB, UB);
- LoopGuard->setName("polly.loop_guard");
- Builder.CreateCondBr(LoopGuard, PreHeaderBB, ExitBB);
+ // GuardBB
+ Builder.SetInsertPoint(GuardBB);
+ Value *LoopGuard;
+ LoopGuard = Builder.CreateICmp(Predicate, LB, UB);
+ LoopGuard->setName("polly.loop_guard");
+ Builder.CreateCondBr(LoopGuard, PreHeaderBB, ExitBB);
+ DT.addNewBlock(PreHeaderBB, GuardBB);
+ } else {
+ BeforeBB->getTerminator()->setSuccessor(0, PreHeaderBB);
+ DT.addNewBlock(PreHeaderBB, BeforeBB);
+ }
// PreHeaderBB
- DT.addNewBlock(PreHeaderBB, GuardBB);
Builder.SetInsertPoint(PreHeaderBB);
Builder.CreateBr(HeaderBB);
@@ -120,7 +124,10 @@
LoopCondition->setName("polly.loop_cond");
Builder.CreateCondBr(LoopCondition, HeaderBB, ExitBB);
IV->addIncoming(IncrementedIV, HeaderBB);
- DT.changeImmediateDominator(ExitBB, GuardBB);
+ if (GuardBB)
+ DT.changeImmediateDominator(ExitBB, GuardBB);
+ else
+ DT.changeImmediateDominator(ExitBB, BeforeBB);
// The loop body should be added here.
Builder.SetInsertPoint(HeaderBB->getFirstNonPHI());
@@ -322,7 +329,7 @@
Builder.SetInsertPoint(--Builder.GetInsertPoint());
LoopInfo &LI = P->getAnalysis<LoopInfo>();
IV = createLoop(LowerBound, UpperBound, Stride, Builder, P, LI, DT, AfterBB,
- ICmpInst::ICMP_SLE);
+ ICmpInst::ICMP_SLE, nullptr, true, /* UseGuard */ false);
BasicBlock::iterator LoopBody = Builder.GetInsertPoint();
Builder.SetInsertPoint(AfterBB->begin());
diff --git a/polly/test/Isl/CodeGen/no_guard_bb.ll b/polly/test/Isl/CodeGen/no_guard_bb.ll
new file mode 100644
index 0000000..3b51c77
--- /dev/null
+++ b/polly/test/Isl/CodeGen/no_guard_bb.ll
@@ -0,0 +1,33 @@
+; RUN: opt %loadPolly -polly-codegen-isl -S < %s | FileCheck %s
+;
+; CHECK-NOT: br i1 true, label %polly.{{.*}}, label %polly.{{.*}}
+;
+; void jd(int *A) {
+; for (int i = 0; i < 1024; i++)
+; A[i] = i;
+; }
+;
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+define void @jd(i32* %A) {
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.inc, %entry
+ %indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ 0, %entry ]
+ %exitcond = icmp ne i64 %indvars.iv, 1024
+ br i1 %exitcond, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv
+ %tmp = trunc i64 %indvars.iv to i32
+ store i32 %tmp, i32* %arrayidx, align 4
+ br label %for.inc
+
+for.inc: ; preds = %for.body
+ %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}