Revert [LoopPeel] Remove known trip count restriction when peeling last. #140792 #142943
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…rds." This reverts commit 3a8b488.
…peeling last. (llvm#140792)"" This commit causes a crash when compiling valid code. See llvm#142830 This reverts commit f98bdd9.
|
@llvm/pr-subscribers-llvm-transforms Author: Alexander Kornienko (alexfh) ChangesRevert [LoopPeel] Remove known trip count restriction when peeling last. #140792 The commit causes a crash in LLVM. See llvm#142830
Patch is 33.05 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/142943.diff 8 Files Affected:
diff --git a/llvm/include/llvm/Transforms/Utils/LoopPeel.h b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
index 49dbc9aa1f2a9..dd59a9c766e45 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopPeel.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
@@ -44,8 +44,7 @@ gatherPeelingPreferences(Loop *L, ScalarEvolution &SE,
void computePeelCount(Loop *L, unsigned LoopSize,
TargetTransformInfo::PeelingPreferences &PP,
unsigned TripCount, DominatorTree &DT,
- ScalarEvolution &SE, const TargetTransformInfo &TTI,
- AssumptionCache *AC = nullptr,
+ ScalarEvolution &SE, AssumptionCache *AC = nullptr,
unsigned Threshold = UINT_MAX);
} // end namespace llvm
diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
index afa7abfea419e..0b9fee5727c6f 100644
--- a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
@@ -1014,7 +1014,7 @@ bool llvm::computeUnrollCount(
}
// 5th priority is loop peeling.
- computePeelCount(L, LoopSize, PP, TripCount, DT, SE, TTI, AC, UP.Threshold);
+ computePeelCount(L, LoopSize, PP, TripCount, DT, SE, AC, UP.Threshold);
if (PP.PeelCount) {
UP.Runtime = false;
UP.Count = 1;
diff --git a/llvm/lib/Transforms/Utils/LoopPeel.cpp b/llvm/lib/Transforms/Utils/LoopPeel.cpp
index bd025fddd0cf7..f348d24ec24fb 100644
--- a/llvm/lib/Transforms/Utils/LoopPeel.cpp
+++ b/llvm/lib/Transforms/Utils/LoopPeel.cpp
@@ -38,7 +38,6 @@
#include "llvm/Transforms/Utils/Cloning.h"
#include "llvm/Transforms/Utils/LoopSimplify.h"
#include "llvm/Transforms/Utils/LoopUtils.h"
-#include "llvm/Transforms/Utils/ScalarEvolutionExpander.h"
#include "llvm/Transforms/Utils/ValueMapper.h"
#include <algorithm>
#include <cassert>
@@ -331,7 +330,11 @@ static unsigned peelToTurnInvariantLoadsDerefencebale(Loop &L,
bool llvm::canPeelLastIteration(const Loop &L, ScalarEvolution &SE) {
const SCEV *BTC = SE.getBackedgeTakenCount(&L);
- if (isa<SCEVCouldNotCompute>(BTC))
+ // The loop must execute at least 2 iterations to guarantee that peeled
+ // iteration executes.
+ // TODO: Add checks during codegen.
+ if (isa<SCEVCouldNotCompute>(BTC) ||
+ !SE.isKnownPredicate(CmpInst::ICMP_UGT, BTC, SE.getZero(BTC->getType())))
return false;
// Check if the exit condition of the loop can be adjusted by the peeling
@@ -351,7 +354,6 @@ bool llvm::canPeelLastIteration(const Loop &L, ScalarEvolution &SE) {
m_BasicBlock(Succ1), m_BasicBlock(Succ2))) &&
((Pred == CmpInst::ICMP_EQ && Succ2 == L.getHeader()) ||
(Pred == CmpInst::ICMP_NE && Succ1 == L.getHeader())) &&
- Bound->getType()->isIntegerTy() &&
SE.isLoopInvariant(SE.getSCEV(Bound), &L) &&
match(SE.getSCEV(Inc),
m_scev_AffineAddRec(m_SCEV(), m_scev_One(), m_SpecificLoop(&L)));
@@ -362,18 +364,12 @@ bool llvm::canPeelLastIteration(const Loop &L, ScalarEvolution &SE) {
/// is known at the second-to-last.
static bool shouldPeelLastIteration(Loop &L, CmpPredicate Pred,
const SCEVAddRecExpr *LeftAR,
- const SCEV *RightSCEV, ScalarEvolution &SE,
- const TargetTransformInfo &TTI) {
+ const SCEV *RightSCEV,
+ ScalarEvolution &SE) {
if (!canPeelLastIteration(L, SE))
return false;
const SCEV *BTC = SE.getBackedgeTakenCount(&L);
- SCEVExpander Expander(SE, L.getHeader()->getDataLayout(), "loop-peel");
- if (!SE.isKnownNonZero(BTC) &&
- Expander.isHighCostExpansion(BTC, &L, SCEVCheapExpansionBudget, &TTI,
- L.getLoopPredecessor()->getTerminator()))
- return false;
-
const SCEV *ValAtLastIter = LeftAR->evaluateAtIteration(BTC, SE);
const SCEV *ValAtSecondToLastIter = LeftAR->evaluateAtIteration(
SE.getMinusSCEV(BTC, SE.getOne(BTC->getType())), SE);
@@ -395,8 +391,7 @@ static bool shouldPeelLastIteration(Loop &L, CmpPredicate Pred,
// ..
// }
static std::pair<unsigned, unsigned>
-countToEliminateCompares(Loop &L, unsigned MaxPeelCount, ScalarEvolution &SE,
- const TargetTransformInfo &TTI) {
+countToEliminateCompares(Loop &L, unsigned MaxPeelCount, ScalarEvolution &SE) {
assert(L.isLoopSimplifyForm() && "Loop needs to be in loop simplify form");
unsigned DesiredPeelCount = 0;
unsigned DesiredPeelCountLast = 0;
@@ -484,7 +479,7 @@ countToEliminateCompares(Loop &L, unsigned MaxPeelCount, ScalarEvolution &SE,
const SCEV *Step = LeftAR->getStepRecurrence(SE);
if (!PeelWhilePredicateIsKnown(NewPeelCount, IterVal, RightSCEV, Step,
Pred)) {
- if (shouldPeelLastIteration(L, Pred, LeftAR, RightSCEV, SE, TTI))
+ if (shouldPeelLastIteration(L, Pred, LeftAR, RightSCEV, SE))
DesiredPeelCountLast = 1;
return;
}
@@ -598,8 +593,8 @@ static bool violatesLegacyMultiExitLoopCheck(Loop *L) {
void llvm::computePeelCount(Loop *L, unsigned LoopSize,
TargetTransformInfo::PeelingPreferences &PP,
unsigned TripCount, DominatorTree &DT,
- ScalarEvolution &SE, const TargetTransformInfo &TTI,
- AssumptionCache *AC, unsigned Threshold) {
+ ScalarEvolution &SE, AssumptionCache *AC,
+ unsigned Threshold) {
assert(LoopSize > 0 && "Zero loop size is not allowed!");
// Save the PP.PeelCount value set by the target in
// TTI.getPeelingPreferences or by the flag -unroll-peel-count.
@@ -661,7 +656,7 @@ void llvm::computePeelCount(Loop *L, unsigned LoopSize,
}
const auto &[CountToEliminateCmps, CountToEliminateCmpsLast] =
- countToEliminateCompares(*L, MaxPeelCount, SE, TTI);
+ countToEliminateCompares(*L, MaxPeelCount, SE);
DesiredPeelCount = std::max(DesiredPeelCount, CountToEliminateCmps);
if (DesiredPeelCount == 0)
@@ -827,7 +822,7 @@ static void initBranchWeights(DenseMap<Instruction *, WeightInfo> &WeightInfos,
/// instructions in the last peeled-off iteration.
static void cloneLoopBlocks(
Loop *L, unsigned IterNumber, bool PeelLast, BasicBlock *InsertTop,
- BasicBlock *InsertBot, BasicBlock *OrigPreHeader,
+ BasicBlock *InsertBot,
SmallVectorImpl<std::pair<BasicBlock *, BasicBlock *>> &ExitEdges,
SmallVectorImpl<BasicBlock *> &NewBlocks, LoopBlocksDFS &LoopBlocks,
ValueToValueMapTy &VMap, ValueToValueMapTy &LVMap, DominatorTree *DT,
@@ -919,22 +914,12 @@ static void cloneLoopBlocks(
// loop iteration. Since this copy is no longer part of the loop, we
// resolve this statically:
if (PeelLast) {
- // For the last iteration, we introduce new phis for each header phi in
- // InsertTop, using the incoming value from the preheader for the original
- // preheader (when skipping the main loop) and the incoming value from the
- // latch for the latch (when continuing from the main loop).
- IRBuilder<> B(InsertTop, InsertTop->getFirstNonPHIIt());
+ // For the last iteration, we use the value from the latch of the original
+ // loop directly.
for (BasicBlock::iterator I = Header->begin(); isa<PHINode>(I); ++I) {
PHINode *NewPHI = cast<PHINode>(VMap[&*I]);
- PHINode *PN = B.CreatePHI(NewPHI->getType(), 2);
+ VMap[&*I] = NewPHI->getIncomingValueForBlock(Latch);
NewPHI->eraseFromParent();
- if (OrigPreHeader)
- PN->addIncoming(cast<PHINode>(&*I)->getIncomingValueForBlock(PreHeader),
- OrigPreHeader);
-
- PN->addIncoming(cast<PHINode>(&*I)->getIncomingValueForBlock(Latch),
- Latch);
- VMap[&*I] = PN;
}
} else {
// For the first iteration, we use the value from the preheader directly.
@@ -1068,7 +1053,7 @@ bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
// Set up all the necessary basic blocks.
BasicBlock *InsertTop;
BasicBlock *InsertBot;
- BasicBlock *NewPreHeader = nullptr;
+ BasicBlock *NewPreHeader;
DenseMap<Instruction *, Value *> ExitValues;
if (PeelLast) {
// It is convenient to split the single exit block from the latch the
@@ -1099,34 +1084,11 @@ bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
for (PHINode &P : Exit->phis())
ExitValues[&P] = P.getIncomingValueForBlock(Latch);
- const SCEV *BTC = SE->getBackedgeTakenCount(L);
-
InsertTop = SplitEdge(Latch, Exit, &DT, LI);
InsertBot = SplitBlock(InsertTop, InsertTop->getTerminator(), &DT, LI);
InsertTop->setName(Exit->getName() + ".peel.begin");
InsertBot->setName(Exit->getName() + ".peel.next");
- NewPreHeader = nullptr;
-
- // If the original loop may only execute a single iteration we need to
- // insert a trip count check and skip the original loop with the last
- // iteration peeled off if necessary.
- if (!SE->isKnownNonZero(BTC)) {
- NewPreHeader = SplitEdge(PreHeader, Header, &DT, LI);
- SCEVExpander Expander(*SE, Latch->getDataLayout(), "loop-peel");
-
- BranchInst *PreHeaderBR = cast<BranchInst>(PreHeader->getTerminator());
- Value *BTCValue =
- Expander.expandCodeFor(BTC, BTC->getType(), PreHeaderBR);
- IRBuilder<> B(PreHeaderBR);
- Value *Cond =
- B.CreateICmpNE(BTCValue, ConstantInt::get(BTCValue->getType(), 0));
- B.CreateCondBr(Cond, NewPreHeader, InsertTop);
- PreHeaderBR->eraseFromParent();
-
- // PreHeader now dominates InsertTop.
- DT.changeImmediateDominator(InsertTop, PreHeader);
- }
} else {
// It is convenient to split the preheader into 3 parts - two blocks to
// anchor the peeled copy of the loop body, and a new preheader for the
@@ -1200,9 +1162,8 @@ bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
for (unsigned Iter = 0; Iter < PeelCount; ++Iter) {
SmallVector<BasicBlock *, 8> NewBlocks;
- cloneLoopBlocks(L, Iter, PeelLast, InsertTop, InsertBot,
- NewPreHeader ? PreHeader : nullptr, ExitEdges, NewBlocks,
- LoopBlocks, VMap, LVMap, &DT, LI,
+ cloneLoopBlocks(L, Iter, PeelLast, InsertTop, InsertBot, ExitEdges,
+ NewBlocks, LoopBlocks, VMap, LVMap, &DT, LI,
LoopLocalNoAliasDeclScopes, *SE);
// Remap to use values from the current iteration instead of the
@@ -1255,11 +1216,9 @@ bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
if (PeelLast) {
// Now adjust users of the original exit values by replacing them with the
- // exit value from the peeled iteration and remove them.
- for (const auto &[P, E] : ExitValues) {
+ // exit value from the peeled iteration.
+ for (const auto &[P, E] : ExitValues)
P->replaceAllUsesWith(isa<Constant>(E) ? E : &*VMap.lookup(E));
- P->eraseFromParent();
- }
formLCSSA(*L, DT, LI, SE);
} else {
// Now adjust the phi nodes in the loop header to get their initial values
diff --git a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-expansion-cost.ll b/llvm/test/Transforms/LoopUnroll/peel-last-iteration-expansion-cost.ll
index f3910f9bfc399..342c70170fef3 100644
--- a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-expansion-cost.ll
+++ b/llvm/test/Transforms/LoopUnroll/peel-last-iteration-expansion-cost.ll
@@ -25,41 +25,17 @@ define i32 @test_expansion_cost_2(i32 %start, i32 %end) {
; BUDGET3-SAME: i32 [[START:%.*]], i32 [[END:%.*]]) {
; BUDGET3-NEXT: [[ENTRY:.*]]:
; BUDGET3-NEXT: [[SUB:%.*]] = add i32 [[END]], -1
-; BUDGET3-NEXT: [[TMP0:%.*]] = sub i32 [[SUB]], [[START]]
-; BUDGET3-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
-; BUDGET3-NEXT: br i1 [[TMP1]], label %[[ENTRY_SPLIT:.*]], label %[[EXIT_PEEL_BEGIN:.*]]
-; BUDGET3: [[ENTRY_SPLIT]]:
; BUDGET3-NEXT: br label %[[LOOP_HEADER:.*]]
; BUDGET3: [[LOOP_HEADER]]:
-; BUDGET3-NEXT: [[IV:%.*]] = phi i32 [ [[START]], %[[ENTRY_SPLIT]] ], [ [[IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ]
-; BUDGET3-NEXT: [[C:%.*]] = icmp eq i32 [[IV]], [[SUB]]
-; BUDGET3-NEXT: br i1 [[C]], label %[[THEN:.*]], label %[[LOOP_LATCH]]
+; BUDGET3-NEXT: [[TMP3:%.*]] = phi i32 [ [[START]], %[[ENTRY]] ], [ [[IV_NEXT_PEEL:%.*]], %[[LOOP_LATCH:.*]] ]
+; BUDGET3-NEXT: [[C_PEEL:%.*]] = icmp eq i32 [[TMP3]], [[SUB]]
+; BUDGET3-NEXT: br i1 [[C_PEEL]], label %[[THEN:.*]], label %[[LOOP_LATCH]]
; BUDGET3: [[THEN]]:
; BUDGET3-NEXT: br label %[[LOOP_LATCH]]
; BUDGET3: [[LOOP_LATCH]]:
-; BUDGET3-NEXT: [[IV_NEXT]] = add nsw i32 [[IV]], 1
-; BUDGET3-NEXT: [[TMP2:%.*]] = sub i32 [[END]], 1
-; BUDGET3-NEXT: [[EC:%.*]] = icmp eq i32 [[IV_NEXT]], [[TMP2]]
-; BUDGET3-NEXT: br i1 [[EC]], label %[[EXIT_PEEL_BEGIN_LOOPEXIT:.*]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP0:![0-9]+]]
-; BUDGET3: [[EXIT_PEEL_BEGIN_LOOPEXIT]]:
-; BUDGET3-NEXT: [[DOTPH:%.*]] = phi i32 [ [[IV_NEXT]], %[[LOOP_LATCH]] ]
-; BUDGET3-NEXT: br label %[[EXIT_PEEL_BEGIN]]
-; BUDGET3: [[EXIT_PEEL_BEGIN]]:
-; BUDGET3-NEXT: [[TMP3:%.*]] = phi i32 [ [[START]], %[[ENTRY]] ], [ [[DOTPH]], %[[EXIT_PEEL_BEGIN_LOOPEXIT]] ]
-; BUDGET3-NEXT: br label %[[LOOP_HEADER_PEEL:.*]]
-; BUDGET3: [[LOOP_HEADER_PEEL]]:
-; BUDGET3-NEXT: [[C_PEEL:%.*]] = icmp eq i32 [[TMP3]], [[SUB]]
-; BUDGET3-NEXT: br i1 [[C_PEEL]], label %[[THEN_PEEL:.*]], label %[[LOOP_LATCH_PEEL:.*]]
-; BUDGET3: [[THEN_PEEL]]:
-; BUDGET3-NEXT: br label %[[LOOP_LATCH_PEEL]]
-; BUDGET3: [[LOOP_LATCH_PEEL]]:
-; BUDGET3-NEXT: [[IV_NEXT_PEEL:%.*]] = add nsw i32 [[TMP3]], 1
+; BUDGET3-NEXT: [[IV_NEXT_PEEL]] = add nsw i32 [[TMP3]], 1
; BUDGET3-NEXT: [[EC_PEEL:%.*]] = icmp eq i32 [[IV_NEXT_PEEL]], [[END]]
-; BUDGET3-NEXT: br i1 [[EC_PEEL]], label %[[EXIT_PEEL_NEXT:.*]], label %[[EXIT_PEEL_NEXT]]
-; BUDGET3: [[EXIT_PEEL_NEXT]]:
-; BUDGET3-NEXT: br label %[[LOOP_HEADER_PEEL_NEXT:.*]]
-; BUDGET3: [[LOOP_HEADER_PEEL_NEXT]]:
-; BUDGET3-NEXT: br label %[[EXIT:.*]]
+; BUDGET3-NEXT: br i1 [[EC_PEEL]], label %[[EXIT:.*]], label %[[LOOP_HEADER]]
; BUDGET3: [[EXIT]]:
; BUDGET3-NEXT: ret i32 0
;
@@ -83,7 +59,3 @@ loop.latch:
exit:
ret i32 0
}
-;.
-; BUDGET3: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]]}
-; BUDGET3: [[META1]] = !{!"llvm.loop.peeled.count", i32 1}
-;.
diff --git a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-constant-trip-count.ll b/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-constant-trip-count.ll
index 5c9b2fa6a9d60..f1290069bda0c 100644
--- a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-constant-trip-count.ll
+++ b/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-constant-trip-count.ll
@@ -13,6 +13,7 @@ define i64 @peel_single_block_loop_iv_step_1() {
; CHECK-NEXT: br i1 [[EC1]], label %[[LOOP]], label %[[EXIT_PEEL_BEGIN:.*]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[EXIT_PEEL_BEGIN]]:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT1]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i64 [ [[IV1]], %[[LOOP]] ]
; CHECK-NEXT: br label %[[LOOP_PEEL:.*]]
; CHECK: [[LOOP_PEEL]]:
; CHECK-NEXT: [[CMP18_NOT:%.*]] = icmp eq i64 [[IV]], 63
@@ -91,6 +92,7 @@ define i64 @peel_single_block_loop_iv_step_1_eq_pred() {
; CHECK-NEXT: br i1 [[CMP_PEEL]], label %[[EXIT_PEEL_BEGIN:.*]], label %[[LOOP]], !llvm.loop [[LOOP2:![0-9]+]]
; CHECK: [[EXIT_PEEL_BEGIN]]:
; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i64 [ [[IV_LCSSA]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV_LCSSA1:%.*]] = phi i64 [ [[IV]], %[[LOOP]] ]
; CHECK-NEXT: br label %[[LOOP_PEEL:.*]]
; CHECK: [[LOOP_PEEL]]:
; CHECK-NEXT: [[CMP_PEEL1:%.*]] = icmp eq i64 [[IV_NEXT_LCSSA]], 63
@@ -169,6 +171,7 @@ define i64 @peel_single_block_loop_iv_step_1_nested_loop() {
; CHECK-NEXT: br i1 [[EC]], label %[[LOOP]], label %[[OUTER_LATCH_PEEL_BEGIN:.*]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[OUTER_LATCH_PEEL_BEGIN]]:
; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i64 [ [[IV_NEXT]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i64 [ [[IV]], %[[LOOP]] ]
; CHECK-NEXT: br label %[[LOOP_PEEL:.*]]
; CHECK: [[LOOP_PEEL]]:
; CHECK-NEXT: [[CMP_PEEL:%.*]] = icmp eq i64 [[IV_NEXT_LCSSA]], 63
@@ -234,6 +237,7 @@ define i64 @peel_multi_block_loop_iv_step_1() {
; CHECK-NEXT: br i1 [[EC]], label %[[LOOP]], label %[[EXIT_PEEL_BEGIN:.*]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[EXIT_PEEL_BEGIN]]:
; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i64 [ [[IV_NEXT]], %[[LATCH]] ]
+; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i64 [ [[IV]], %[[LATCH]] ]
; CHECK-NEXT: br label %[[LOOP_PEEL:.*]]
; CHECK: [[LOOP_PEEL]]:
; CHECK-NEXT: [[CMP_PEEL:%.*]] = icmp eq i64 [[IV_NEXT_LCSSA]], 63
@@ -361,6 +365,7 @@ define i64 @peel_single_block_loop_iv_step_1_btc_1() {
; CHECK-NEXT: br i1 false, label %[[LOOP]], label %[[EXIT_PEEL_BEGIN:.*]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: [[EXIT_PEEL_BEGIN]]:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT1]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i64 [ [[IV1]], %[[LOOP]] ]
; CHECK-NEXT: br label %[[LOOP_PEEL:.*]]
; CHECK: [[LOOP_PEEL]]:
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[IV]], 1
@@ -478,8 +483,9 @@ define i32 @peel_loop_with_branch_and_phi_uses(ptr %x, i1 %c) {
; CHECK-NEXT: [[EC1:%.*]] = icmp ne i32 [[IV_NEXT1]], 99
; CHECK-NEXT: br i1 [[EC1]], label %[[LOOP_HEADER]], label %[[LOOPEXIT_PEEL_BEGIN:.*]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: [[LOOPEXIT_PEEL_BEGIN]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT1]], %[[LOOP_LATCH]] ]
; CHECK-NEXT: [[RED:%.*]] = phi i32 [ [[ADD1]], %[[LOOP_LATCH]] ]
+; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT1]], %[[LOOP_LATCH]] ]
+; CHECK-NEXT: [[ADD_LCSSA:%.*]] = phi i32 [ [[ADD1]], %[[LOOP_LATCH]] ]
; CHECK-NEXT: br label %[[LOOP_HEADER_PEEL:.*]]
; CHECK: [[LOOP_HEADER_PEEL]]:
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i32 [[IV]], 99
diff --git a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-guards.ll b/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-guards.ll
deleted file mode 100644
index af07a97131322..0000000000000
--- a/llvm/test/Transforms/LoopUnroll/peel-last-iteration-with-guards.ll
+++ /dev/null
@@ -1,191 +0,0 @@
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt -p loop-unroll -unroll-full-max-count=0 -S %s | FileCheck %s
-
-declare void @foo(i32)
-
-define void @peel_with_guard_known_nonnegative_1(i32 %n) {
-; CHECK-LABEL: define void @peel_with_guard_known_nonnegative_1(
-; CHECK-SAME: i32 [[N:%.*]]) {
-; CHECK-NEXT: [[ENTRY:.*:]]
-; CHECK-NEXT: [[PRECOND:%.*]] = icmp slt i32 [[N]], 0
-; CHECK-NEXT: br i1 [[PRECOND]], label %[[EXIT:.*]], label %[[PH:.*]]
-; CHECK: [[PH]]:
-; CHECK-NEXT: [[N_EXT:%.*]] = zext i32 [[N]] to i64
-; CHECK-NEXT: [[N_1:%.*]] = add i32 [[N]], 1
-; CHECK-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N_1]] to i64
-; CHECK-NEXT: br label %[[LOOP:.*]]
-; CHECK: [[LOOP]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, %[[PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; CHECK-NEXT: [[C:%.*]] = icmp eq i64 [[IV]], [[N_EXT]]
-; CHECK-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[C]], i32 10, i32 20
-; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
-; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[WIDE_TRIP_COUNT]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT_LOOPEXIT:.*]], label %[[LOOP]]
-; CHECK: [[EXIT_LOOPEXIT]]:
-; CHECK-NEXT: br label %[[EXIT]]
-; CHECK: [[EXIT]]:
-; CHECK-NEXT: ret void
-;
-entry:
- %precond = icmp slt i32 %n, 0
- br i1 %precond, label %exit, label %ph
-
-ph:
- %n.ext = zext i32 %n to i64
- %n.1 = add i32 %n, 1
- %wide.trip.count = zext i32 %n.1 to i64
- br label %loop
-
-loop:
- %iv = phi i64 [ 0, %ph ], [ %iv.next, %loop ]
- %c = icmp eq i64 %iv, %n.ext
- %spec.select = select i1 %c, i32 10, i32 20
- %iv.next = add i64 %iv, 1
- %ec = icmp eq i64 %iv.next, %wide.trip.count
- br i1 %ec, label %exit, label %loop
-
-exit:
- ret v...
[truncated]
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Revert [LoopPeel] Remove known trip count restriction when peeling last. #140792
The commit causes a crash in LLVM. See #142830