RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 20:23:11 +01:00
Code Issues Projects Releases Wiki Activity

[LICM] Report successful hoist/sink/promotion

Browse Source 
Differential Revision: https://reviews.llvm.org/D27938

llvm-svn: 291646
This commit is contained in:
Adam Nemet 2017-01-11 04:39:35 +00:00
parent 63cd0c2c97
commit 7d55194036
27 changed files with 161 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
include/llvm/Transforms/Utils/LoopUtils.h
View File

@ -29,6 +29,7 @@ class DataLayout;
class DominatorTree;
class Loop;
class LoopInfo;
class OptimizationRemarkEmitter;
class Pass;
class PredicatedScalarEvolution;
class PredIteratorCache;
@ -404,11 +405,11 @@ bool formLCSSARecursively(Loop &L, DominatorTree &DT, LoopInfo *LI,
/// uses before definitions, allowing us to sink a loop body in one pass without
/// iteration. Takes DomTreeNode, AliasAnalysis, LoopInfo, DominatorTree,
/// DataLayout, TargetLibraryInfo, Loop, AliasSet information for all
/// instructions of the loop and loop safety information as arguments.
/// It returns changed status.
/// instructions of the loop and loop safety information as
/// arguments. Diagnostics is emitted via \p ORE. It returns changed status.
bool sinkRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
TargetLibraryInfo *, Loop *, AliasSetTracker *,
LoopSafetyInfo *);
LoopSafetyInfo *, OptimizationRemarkEmitter *ORE);
/// \brief Walk the specified region of the CFG (defined by all blocks
/// dominated by the specified block, and that are in the current loop) in depth
@ -416,10 +417,11 @@ bool sinkRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
/// before uses, allowing us to hoist a loop body in one pass without iteration.
/// Takes DomTreeNode, AliasAnalysis, LoopInfo, DominatorTree, DataLayout,
/// TargetLibraryInfo, Loop, AliasSet information for all instructions of the
/// loop and loop safety information as arguments. It returns changed status.
/// loop and loop safety information as arguments. Diagnostics is emitted via \p
/// ORE. It returns changed status.
bool hoistRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
TargetLibraryInfo *, Loop *, AliasSetTracker *,
LoopSafetyInfo *);
LoopSafetyInfo *, OptimizationRemarkEmitter *ORE);
/// \brief Try to promote memory values to scalars by sinking stores out of
/// the loop and moving loads to before the loop. We do this by looping over
@ -427,12 +429,14 @@ bool hoistRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
/// loop invariant. It takes AliasSet, Loop exit blocks vector, loop exit blocks
/// insertion point vector, PredIteratorCache, LoopInfo, DominatorTree, Loop,
/// AliasSet information for all instructions of the loop and loop safety
/// information as arguments. It returns changed status.
/// information as arguments. Diagnostics is emitted via \p ORE. It returns
/// changed status.
bool promoteLoopAccessesToScalars(AliasSet &, SmallVectorImpl<BasicBlock *> &,
SmallVectorImpl<Instruction *> &,
PredIteratorCache &, LoopInfo *,
DominatorTree *, const TargetLibraryInfo *,
Loop *, AliasSetTracker *, LoopSafetyInfo *);
Loop *, AliasSetTracker *, LoopSafetyInfo *,
OptimizationRemarkEmitter *);
/// \brief Computes safety information for a loop
/// checks loop body & header for the possibility of may throw

64
lib/Transforms/Scalar/LICM.cpp
View File

@ -43,6 +43,7 @@
#include "llvm/Analysis/LoopPass.h"
#include "llvm/Analysis/LoopPassManager.h"
#include "llvm/Analysis/MemoryBuiltins.h"
#include "llvm/Analysis/OptimizationDiagnosticInfo.h"
#include "llvm/Analysis/ScalarEvolution.h"
#include "llvm/Analysis/ScalarEvolutionAliasAnalysis.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
@ -84,10 +85,12 @@ static bool inSubLoop(BasicBlock *BB, Loop *CurLoop, LoopInfo *LI);
static bool isNotUsedInLoop(const Instruction &I, const Loop *CurLoop,
const LoopSafetyInfo *SafetyInfo);
static bool hoist(Instruction &I, const DominatorTree *DT, const Loop *CurLoop,
const LoopSafetyInfo *SafetyInfo);
const LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE);
static bool sink(Instruction &I, const LoopInfo *LI, const DominatorTree *DT,
const Loop *CurLoop, AliasSetTracker *CurAST,
const LoopSafetyInfo *SafetyInfo);
const LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE);
static bool isSafeToExecuteUnconditionally(const Instruction &Inst,
const DominatorTree *DT,
const Loop *CurLoop,
@ -104,7 +107,8 @@ CloneInstructionInExitBlock(Instruction &I, BasicBlock &ExitBlock, PHINode &PN,
namespace {
struct LoopInvariantCodeMotion {
bool runOnLoop(Loop *L, AliasAnalysis *AA, LoopInfo *LI, DominatorTree *DT,
TargetLibraryInfo *TLI, ScalarEvolution *SE, bool DeleteAST);
TargetLibraryInfo *TLI, ScalarEvolution *SE,
OptimizationRemarkEmitter *ORE, bool DeleteAST);
DenseMap<Loop *, AliasSetTracker *> &getLoopToAliasSetMap() {
return LoopToAliasSetMap;
@ -135,12 +139,16 @@ struct LegacyLICMPass : public LoopPass {
}
auto *SE = getAnalysisIfAvailable<ScalarEvolutionWrapperPass>();
// For the old PM, we can't use OptimizationRemarkEmitter as an analysis
// pass. Function analyses need to be preserved across loop transformations
// but ORE cannot be preserved (see comment before the pass definition).
OptimizationRemarkEmitter ORE(L->getHeader()->getParent());
return LICM.runOnLoop(L,
&getAnalysis<AAResultsWrapperPass>().getAAResults(),
&getAnalysis<LoopInfoWrapperPass>().getLoopInfo(),
&getAnalysis<DominatorTreeWrapperPass>().getDomTree(),
&getAnalysis<TargetLibraryInfoWrapperPass>().getTLI(),
SE ? &SE->getSE() : nullptr, false);
SE ? &SE->getSE() : nullptr, &ORE, false);
}
/// This transformation requires natural loop information & requires that
@ -186,11 +194,13 @@ PreservedAnalyses LICMPass::run(Loop &L, LoopAnalysisManager &AM) {
auto *DT = FAM.getCachedResult<DominatorTreeAnalysis>(*F);
auto *TLI = FAM.getCachedResult<TargetLibraryAnalysis>(*F);
auto *SE = FAM.getCachedResult<ScalarEvolutionAnalysis>(*F);
assert((AA && LI && DT && TLI && SE) && "Analyses for LICM not available");
auto *ORE = FAM.getCachedResult<OptimizationRemarkEmitterAnalysis>(*F);
assert((AA && LI && DT && TLI && SE && ORE) &&
"Analyses for LICM not available");
LoopInvariantCodeMotion LICM;
if (!LICM.runOnLoop(&L, AA, LI, DT, TLI, SE, true))
if (!LICM.runOnLoop(&L, AA, LI, DT, TLI, SE, ORE, true))
return PreservedAnalyses::all();
// FIXME: There is no setPreservesCFG in the new PM. When that becomes
@ -217,7 +227,9 @@ Pass *llvm::createLICMPass() { return new LegacyLICMPass(); }
bool LoopInvariantCodeMotion::runOnLoop(Loop *L, AliasAnalysis *AA,
LoopInfo *LI, DominatorTree *DT,
TargetLibraryInfo *TLI,
ScalarEvolution *SE, bool DeleteAST) {
ScalarEvolution *SE,
OptimizationRemarkEmitter *ORE,
bool DeleteAST) {
bool Changed = false;
assert(L->isLCSSAForm(*DT) && "Loop is not in LCSSA form.");
@ -243,10 +255,10 @@ bool LoopInvariantCodeMotion::runOnLoop(Loop *L, AliasAnalysis *AA,
//
if (L->hasDedicatedExits())
Changed |= sinkRegion(DT->getNode(L->getHeader()), AA, LI, DT, TLI, L,
CurAST, &SafetyInfo);
CurAST, &SafetyInfo, ORE);
if (Preheader)
Changed |= hoistRegion(DT->getNode(L->getHeader()), AA, LI, DT, TLI, L,
CurAST, &SafetyInfo);
CurAST, &SafetyInfo, ORE);
// Now that all loop invariants have been removed from the loop, promote any
// memory references to scalars that we can.
@ -279,7 +291,7 @@ bool LoopInvariantCodeMotion::runOnLoop(Loop *L, AliasAnalysis *AA,
for (AliasSet &AS : *CurAST)
Promoted |=
promoteLoopAccessesToScalars(AS, ExitBlocks, InsertPts, PIC, LI, DT,
TLI, L, CurAST, &SafetyInfo);
TLI, L, CurAST, &SafetyInfo, ORE);
// Once we have promoted values across the loop body we have to
// recursively reform LCSSA as any nested loop may now have values defined
@ -320,7 +332,8 @@ bool LoopInvariantCodeMotion::runOnLoop(Loop *L, AliasAnalysis *AA,
///
bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
DominatorTree *DT, TargetLibraryInfo *TLI, Loop *CurLoop,
AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo) {
AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE) {
// Verify inputs.
assert(N != nullptr && AA != nullptr && LI != nullptr && DT != nullptr &&
@ -336,7 +349,8 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
bool Changed = false;
const std::vector<DomTreeNode *> &Children = N->getChildren();
for (DomTreeNode *Child : Children)
Changed |= sinkRegion(Child, AA, LI, DT, TLI, CurLoop, CurAST, SafetyInfo);
Changed |=
sinkRegion(Child, AA, LI, DT, TLI, CurLoop, CurAST, SafetyInfo, ORE);
// Only need to process the contents of this block if it is not part of a
// subloop (which would already have been processed).
@ -365,7 +379,7 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
if (isNotUsedInLoop(I, CurLoop, SafetyInfo) &&
canSinkOrHoistInst(I, AA, DT, CurLoop, CurAST, SafetyInfo)) {
++II;
Changed |= sink(I, LI, DT, CurLoop, CurAST, SafetyInfo);
Changed |= sink(I, LI, DT, CurLoop, CurAST, SafetyInfo, ORE);
}
}
return Changed;
@ -378,7 +392,8 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
///
bool llvm::hoistRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
DominatorTree *DT, TargetLibraryInfo *TLI, Loop *CurLoop,
AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo) {
AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE) {
// Verify inputs.
assert(N != nullptr && AA != nullptr && LI != nullptr && DT != nullptr &&
CurLoop != nullptr && CurAST != nullptr && SafetyInfo != nullptr &&
@ -421,12 +436,13 @@ bool llvm::hoistRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
isSafeToExecuteUnconditionally(
I, DT, CurLoop, SafetyInfo,
CurLoop->getLoopPreheader()->getTerminator()))
Changed |= hoist(I, DT, CurLoop, SafetyInfo);
Changed |= hoist(I, DT, CurLoop, SafetyInfo, ORE);
}
const std::vector<DomTreeNode *> &Children = N->getChildren();
for (DomTreeNode *Child : Children)
Changed |= hoistRegion(Child, AA, LI, DT, TLI, CurLoop, CurAST, SafetyInfo);
Changed |=
hoistRegion(Child, AA, LI, DT, TLI, CurLoop, CurAST, SafetyInfo, ORE);
return Changed;
}
@ -680,8 +696,11 @@ CloneInstructionInExitBlock(Instruction &I, BasicBlock &ExitBlock, PHINode &PN,
///
static bool sink(Instruction &I, const LoopInfo *LI, const DominatorTree *DT,
const Loop *CurLoop, AliasSetTracker *CurAST,
const LoopSafetyInfo *SafetyInfo) {
const LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE) {
DEBUG(dbgs() << "LICM sinking instruction: " << I << "\n");
ORE->emit(OptimizationRemark(DEBUG_TYPE, "InstSunk", &I)
<< "sinking " << ore::NV("Inst", &I));
bool Changed = false;
if (isa<LoadInst>(I))
++NumMovedLoads;
@ -748,10 +767,13 @@ static bool sink(Instruction &I, const LoopInfo *LI, const DominatorTree *DT,
/// is safe to hoist, this instruction is called to do the dirty work.
///
static bool hoist(Instruction &I, const DominatorTree *DT, const Loop *CurLoop,
const LoopSafetyInfo *SafetyInfo) {
const LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE) {
auto *Preheader = CurLoop->getLoopPreheader();
DEBUG(dbgs() << "LICM hoisting to " << Preheader->getName() << ": " << I
<< "\n");
ORE->emit(OptimizationRemark(DEBUG_TYPE, "Hoisted", &I)
<< "hosting " << ore::NV("Inst", &I));
// Metadata can be dependent on conditions we are hoisting above.
// Conservatively strip all metadata on the instruction unless we were
@ -882,7 +904,8 @@ bool llvm::promoteLoopAccessesToScalars(
AliasSet &AS, SmallVectorImpl<BasicBlock *> &ExitBlocks,
SmallVectorImpl<Instruction *> &InsertPts, PredIteratorCache &PIC,
LoopInfo *LI, DominatorTree *DT, const TargetLibraryInfo *TLI,
Loop *CurLoop, AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo) {
Loop *CurLoop, AliasSetTracker *CurAST, LoopSafetyInfo *SafetyInfo,
OptimizationRemarkEmitter *ORE) {
// Verify inputs.
assert(LI != nullptr && DT != nullptr && CurLoop != nullptr &&
CurAST != nullptr && SafetyInfo != nullptr &&
@ -1074,6 +1097,9 @@ bool llvm::promoteLoopAccessesToScalars(
// Otherwise, this is safe to promote, lets do it!
DEBUG(dbgs() << "LICM: Promoting value stored to in loop: " << *SomePtr
<< '\n');
ORE->emit(
OptimizationRemark(DEBUG_TYPE, "PromoteLoopAccessesToScalar", LoopUses[0])
<< "Moving accesses to memory location out of the loop");
++NumPromoted;
// Grab a debug location for the inserted loads/stores; given that the

2
test/Transforms/LICM/argmemonly-call.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -basicaa -licm %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
declare i32 @foo() readonly argmemonly nounwind
declare i32 @foo2() readonly nounwind
declare i32 @bar(i32* %loc2) readonly argmemonly nounwind

2
test/Transforms/LICM/assume.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -licm -basicaa < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
define void @f_0(i1 %p) nounwind ssp {
; CHECK-LABEL: @f_0(

2
test/Transforms/LICM/atomics.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -S -basicaa -licm | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
; Check that we can hoist unordered loads
define i32 @test1(i32* nocapture %y) nounwind uwtable ssp {

2
test/Transforms/LICM/basictest.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -licm | llvm-dis
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s | llvm-dis
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s | llvm-dis
define void @testfunc(i32 %i) {
; <label>:0

2
test/Transforms/LICM/constexpr.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -S -basicaa -licm | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
; This fixes PR22460
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"

2
test/Transforms/LICM/crash.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -licm -disable-output < %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -disable-output < %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -disable-output < %s
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
target triple = "x86_64-apple-darwin10.0.0"

2
test/Transforms/LICM/debug-value.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -licm -basicaa < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
define void @dgefa() nounwind ssp {
entry:

2
test/Transforms/LICM/extra-copies.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -licm -S | FileCheck %s
; RUN: opt -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
; PR19835
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

2
test/Transforms/LICM/funclet.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -licm -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s -S | FileCheck %s
target datalayout = "e-m:x-p:32:32-i64:64-f80:32-n8:16:32-a:0:32-S32"
target triple = "i386-pc-windows-msvc18.0.0"

2
test/Transforms/LICM/hoist-bitcast-load.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -basicaa -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='loop-simplify,require<aa>,require<targetir>,require<scalar-evolution>,loop(simplify-cfg,licm)' -S < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='loop-simplify,require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(simplify-cfg,licm)' -S < %s | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

2
test/Transforms/LICM/hoist-deref-load.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -basicaa -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='loop-simplify,require<aa>,require<targetir>,require<scalar-evolution>,loop(simplify-cfg,licm)' -S < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='loop-simplify,require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(simplify-cfg,licm)' -S < %s | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

2
test/Transforms/LICM/hoist-nounwind.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -basicaa -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='lcssa,require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

2
test/Transforms/LICM/hoist-round.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
target datalayout = "E-m:e-p:32:32-i8:8:8-i16:16:16-i64:32:32-f64:32:32-v64:32:32-v128:32:32-a0:0:32-n32"

2
test/Transforms/LICM/hoisting.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -licm -S | FileCheck %s
; RUN: opt -lcssa %s | opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S | FileCheck %s
; RUN: opt -lcssa %s | opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S | FileCheck %s
@X = global i32 0 ; <i32*> [#uses=1]

2
test/Transforms/LICM/lcssa-ssa-promoter.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -basicaa -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s| FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s| FileCheck %s
;
; Manually validate LCSSA form is preserved even after SSAUpdater is used to
; promote things in the loop bodies.

2
test/Transforms/LICM/no-preheader-test.ll
View File

@ -1,6 +1,6 @@
; Test that LICM works when there is not a loop-preheader
; RUN: opt < %s -licm | llvm-dis
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' < %s | llvm-dis
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' < %s | llvm-dis
define void @testfunc(i32 %i.s, i1 %ifcond) {
br i1 %ifcond, label %Then, label %Else

81
test/Transforms/LICM/opt-remarks.ll Normal file
View File

@ -0,0 +1,81 @@
; RUN: opt < %s -licm -pass-remarks=licm -o /dev/null 2>&1 | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' %s -o /dev/null -pass-remarks=licm 2>&1 | FileCheck %s
target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
define void @hoist(i32* %array, i32* noalias %p) {
Entry:
br label %Loop
Loop:
%j = phi i32 [ 0, %Entry ], [ %Next, %Loop ]
%addr = getelementptr i32, i32* %array, i32 %j
%a = load i32, i32* %addr
; CHECK: remark: /tmp/kk.c:2:20: hosting load
%b = load i32, i32* %p, !dbg !8
%a2 = add i32 %a, %b
store i32 %a2, i32* %addr
%Next = add i32 %j, 1
%cond = icmp eq i32 %Next, 0
br i1 %cond, label %Out, label %Loop
Out:
ret void
}
define i32 @sink(i32* %array, i32* noalias %p, i32 %b) {
Entry:
br label %Loop
Loop:
%j = phi i32 [ 0, %Entry ], [ %Next, %Loop ]
%addr = getelementptr i32, i32* %array, i32 %j
%a = load i32, i32* %addr
%a2 = add i32 %a, %b
store i32 %a2, i32* %addr
; CHECK: remark: /tmp/kk.c:2:21: sinking add
%a3 = add i32 %a, 1, !dbg !9
%Next = add i32 %j, 1
%cond = icmp eq i32 %Next, 0
br i1 %cond, label %Out, label %Loop
Out:
%a4 = phi i32 [ %a3, %Loop ]
ret i32 %a4
}
define void @promote(i32* %array, i32* noalias %p) {
Entry:
br label %Loop
Loop:
%j = phi i32 [ 0, %Entry ], [ %Next, %Loop ]
%addr = getelementptr i32, i32* %array, i32 %j
%a = load i32, i32* %addr
%b = load i32, i32* %p
%a2 = add i32 %a, %b
store i32 %a2, i32* %addr
; CHECK: remark: /tmp/kk.c:2:22: Moving accesses to memory location out of the loop
store i32 %b, i32* %p, !dbg !10
%Next = add i32 %j, 1
%cond = icmp eq i32 %Next, 0
br i1 %cond, label %Out, label %Loop
Out:
ret void
}
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4}
!llvm.ident = !{!5}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 3.9.0 ", isOptimized: true, runtimeVersion: 0, emissionKind: NoDebug, enums: !2)
!1 = !DIFile(filename: "/tmp/kk.c", directory: "/tmp")
!2 = !{}
!3 = !{i32 2, !"Debug Info Version", i32 3}
!4 = !{i32 1, !"PIC Level", i32 2}
!5 = !{!"clang version 3.9.0 "}
!6 = distinct !DISubprogram(name: "success", scope: !1, file: !1, line: 1, type: !7, isLocal: false, isDefinition: true, scopeLine: 1, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !2)
!7 = !DISubroutineType(types: !2)
!8 = !DILocation(line: 2, column: 20, scope: !6)
!9 = !DILocation(line: 2, column: 21, scope: !6)
!10 = !DILocation(line: 2, column: 22, scope: !6)

2
test/Transforms/LICM/preheader-safe.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -licm < %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
declare void @use_nothrow(i64 %a) nounwind
declare void @use(i64 %a)

2
test/Transforms/LICM/promote-order.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -tbaa -basicaa -licm -S < %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
; LICM should keep the stores in their original order when it sinks/promotes them.
; rdar://12045203

2
test/Transforms/LICM/promote-tls.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -tbaa -basicaa -licm -S < %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
; If we can prove a local is thread local, we can insert stores during
; promotion which wouldn't be legal otherwise.

2
test/Transforms/LICM/scalar-promote-memmodel.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -basicaa -licm -S | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
; Make sure we don't hoist a conditionally-executed store out of the loop;
; it would violate the concurrency memory model

2
test/Transforms/LICM/scalar_promote-unwind.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -basicaa -licm -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

2
test/Transforms/LICM/scalar_promote.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -basicaa -tbaa -licm -S | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -aa-pipeline=type-based-aa,basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
@X = global i32 7 ; <i32*> [#uses=4]

2
test/Transforms/LICM/speculate.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -S -licm < %s | FileCheck %s
; RUN: opt -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S %s | FileCheck %s
; RUN: opt -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S %s | FileCheck %s
; UDiv is safe to speculate if the denominator is known non-zero.

2
test/Transforms/LICM/volatile-alias.ll
View File

@ -1,5 +1,5 @@
; RUN: opt -basicaa -sroa -loop-rotate -licm -S < %s | FileCheck %s
; RUN: opt -basicaa -sroa -loop-rotate %s | opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,loop(licm)' -S | FileCheck %s
; RUN: opt -basicaa -sroa -loop-rotate %s | opt -aa-pipeline=basic-aa -passes='require<aa>,require<targetir>,require<scalar-evolution>,require<opt-remark-emit>,loop(licm)' -S | FileCheck %s
; The objects *p and *q are aliased to each other, but even though *q is
; volatile, *p can be considered invariant in the loop. Check if it is moved
; out of the loop.