[llvm] [SCEV] Try to re-use pointer LCSSA phis when expanding SCEVs. (PR #147824)
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Fri Jul 25 03:45:53 PDT 2025
https://github.com/fhahn updated https://github.com/llvm/llvm-project/pull/147824
>From 018afa588610749d53309a8d85751650bd600023 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Fri, 4 Jul 2025 21:52:02 +0100
Subject: [PATCH 1/4] [SCEV] Try to re-use pointer LCSSA phis when expanding
SCEVs.
Generalize the code added in
https://github.com/llvm/llvm-project/pull/147214 to also support
re-using pointer LCSSA phis when expanding integer SCEVs with AddRecs.
A common source of integer AddRecs with pointer bases are runtime checks
emitted by LV based on the distance between 2 pointer AddRecs.
This improves codegen in some cases when vectorizing and prevents
regressions with https://github.com/llvm/llvm-project/pull/142309, which
turns some phis into single-entry ones, which SCEV will look through
now (and expand the whole AddRec), whereas before it would have to treat
the LCSSA phi as SCEVUnknown.
---
.../Utils/ScalarEvolutionExpander.cpp | 23 +++++++++++++++----
.../invalidate-laa-after-versioning.ll | 11 ++++-----
.../reuse-lcssa-phi-scev-expansion.ll | 20 +++++++---------
.../invalidate-laa-after-versioning.ll | 15 +++++-------
4 files changed, 37 insertions(+), 32 deletions(-)
diff --git a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
index ed08c0bfa2e7d..97de12ae4c053 100644
--- a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
+++ b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
@@ -1224,6 +1224,7 @@ Value *SCEVExpander::expandAddRecExprLiterally(const SCEVAddRecExpr *S) {
}
Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
+ Type *STy = S->getType();
const Loop *L = S->getLoop();
BasicBlock *EB = L->getExitBlock();
if (!EB || !EB->getSinglePredecessor() ||
@@ -1231,11 +1232,25 @@ Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
return nullptr;
for (auto &PN : EB->phis()) {
- if (!SE.isSCEVable(PN.getType()) || PN.getType() != S->getType())
+ if (!SE.isSCEVable(PN.getType()))
continue;
- auto *ExitV = SE.getSCEV(&PN);
- if (S == ExitV)
- return &PN;
+ auto *ExitSCEV = SE.getSCEV(&PN);
+ Type *PhiTy = PN.getType();
+ if (STy->isIntegerTy() && PhiTy->isPointerTy())
+ ExitSCEV = SE.getPtrToIntExpr(ExitSCEV, STy);
+ else if (S->getType() != PN.getType())
+ continue;
+ const SCEV *Diff = SE.getMinusSCEV(S, ExitSCEV);
+ if (isa<SCEVCouldNotCompute>(Diff) ||
+ SCEVExprContains(Diff,
+ [](const SCEV *S) { return isa<SCEVAddRecExpr>(S); }))
+ continue;
+
+ Value *DiffV = expand(Diff);
+ Value *BaseV = &PN;
+ if (DiffV->getType()->isIntegerTy() && PhiTy->isPointerTy())
+ BaseV = Builder.CreatePtrToInt(BaseV, DiffV->getType());
+ return Builder.CreateAdd(BaseV, DiffV);
}
return nullptr;
diff --git a/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll b/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
index 10e10653a431d..3ad262bb20910 100644
--- a/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
+++ b/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
@@ -59,19 +59,16 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK-NEXT: [[GEP_5:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_1]], i64 1
; CHECK-NEXT: br label [[INNER_2:%.*]]
; CHECK: inner.2:
-; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], [[INNER_2]] ], [ 0, [[INNER_1_EXIT]] ]
; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[GEP_5]], [[INNER_1_EXIT]] ], [ [[PTR_IV_2_NEXT:%.*]], [[INNER_2]] ]
; CHECK-NEXT: [[PTR_IV_2_NEXT]] = getelementptr inbounds double, ptr [[PTR_IV_2]], i64 1
-; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[INDVAR]], 1
; CHECK-NEXT: br i1 false, label [[INNER_3_LVER_CHECK:%.*]], label [[INNER_2]]
; CHECK: inner.3.lver.check:
-; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i64 [ [[INDVAR]], [[INNER_2]] ]
; CHECK-NEXT: [[LCSSA_PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2]], [[INNER_2]] ]
; CHECK-NEXT: [[GEP_6:%.*]] = getelementptr inbounds double, ptr [[PTR_PHI]], i64 1
; CHECK-NEXT: [[GEP_7:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_2]], i64 1
-; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDVAR_LCSSA]], 3
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 24
-; CHECK-NEXT: [[SCEVGEP3:%.*]] = getelementptr i8, ptr [[LCSSA_PTR_IV_1]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[LCSSA_PTR_IV_2]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 16
+; CHECK-NEXT: [[SCEVGEP3:%.*]] = inttoptr i64 [[TMP1]] to ptr
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[GEP_7]], [[GEP_1]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[PTR_PHI]], [[SCEVGEP3]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
@@ -104,7 +101,7 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK-NEXT: br i1 [[C_2]], label [[OUTER_LATCH_LOOPEXIT4:%.*]], label [[INNER_3]]
; CHECK: outer.latch.loopexit:
; CHECK-NEXT: br label [[OUTER_LATCH]]
-; CHECK: outer.latch.loopexit4:
+; CHECK: outer.latch.loopexit3:
; CHECK-NEXT: br label [[OUTER_LATCH]]
; CHECK: outer.latch:
; CHECK-NEXT: br label [[INNER_1_LVER_CHECK]]
diff --git a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
index 1fc34db26261e..b9b21e986e422 100644
--- a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
+++ b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
@@ -104,27 +104,23 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-LABEL: define void @runtime_checks_ptr_inductions(
; CHECK-SAME: ptr [[DST_1:%.*]], ptr [[DST_2:%.*]], i1 [[C:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*]]:
-; CHECK-NEXT: [[DST_11:%.*]] = ptrtoint ptr [[DST_1]] to i64
; CHECK-NEXT: br label %[[LOOP_1:.*]]
; CHECK: [[LOOP_1]]:
-; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], %[[LOOP_1]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: [[PTR_IV_1:%.*]] = phi ptr [ [[DST_1]], %[[ENTRY]] ], [ [[PTR_IV_1_NEXT:%.*]], %[[LOOP_1]] ]
; CHECK-NEXT: [[CALL:%.*]] = call i32 @val()
; CHECK-NEXT: [[SEL_DST:%.*]] = select i1 [[C]], ptr [[DST_1]], ptr [[DST_2]]
; CHECK-NEXT: [[PTR_IV_1_NEXT]] = getelementptr i8, ptr [[PTR_IV_1]], i64 1
; CHECK-NEXT: [[EC_1:%.*]] = icmp eq i32 [[CALL]], 0
-; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[INDVAR]], 1
; CHECK-NEXT: br i1 [[EC_1]], label %[[LOOP_2_HEADER_PREHEADER:.*]], label %[[LOOP_1]]
; CHECK: [[LOOP_2_HEADER_PREHEADER]]:
-; CHECK-NEXT: [[SEL_DST_LCSSA2:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i64 [ [[INDVAR]], %[[LOOP_1]] ]
+; CHECK-NEXT: [[SEL_DST_LCSSA1:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1]] ]
; CHECK-NEXT: [[SEL_DST_LCSSA:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[SEL_DST_LCSSA23:%.*]] = ptrtoint ptr [[SEL_DST_LCSSA2]] to i64
+; CHECK-NEXT: [[SEL_DST_LCSSA12:%.*]] = ptrtoint ptr [[SEL_DST_LCSSA1]] to i64
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDVAR_LCSSA]], [[DST_11]]
-; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[SEL_DST_LCSSA23]]
+; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[PTR_IV_1_LCSSA]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[SEL_DST_LCSSA12]]
; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP1]], 2
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
@@ -146,13 +142,13 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ 1023, %[[MIDDLE_BLOCK]] ], [ 1, %[[LOOP_2_HEADER_PREHEADER]] ], [ 1, %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA]], %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL6:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL4:%.*]] = phi ptr [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label %[[LOOP_2_HEADER:.*]]
; CHECK: [[LOOP_2_HEADER]]:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[DEC7:%.*]], %[[LOOP_2_LATCH:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL6]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL4]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i32 [[IV]], 1024
; CHECK-NEXT: br i1 [[EC_2]], label %[[EXIT:.*]], label %[[LOOP_2_LATCH]]
; CHECK: [[LOOP_2_LATCH]]:
diff --git a/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll b/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
index 8075314a65b49..858864276c0a0 100644
--- a/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
+++ b/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
@@ -56,19 +56,16 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK-NEXT: [[GEP_5:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_1]], i64 1
; CHECK-NEXT: br label [[INNER_2:%.*]]
; CHECK: inner.2:
-; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], [[INNER_2]] ], [ 0, [[INNER_1_EXIT]] ]
; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[GEP_5]], [[INNER_1_EXIT]] ], [ [[PTR_IV_2_NEXT:%.*]], [[INNER_2]] ]
; CHECK-NEXT: [[PTR_IV_2_NEXT]] = getelementptr inbounds double, ptr [[PTR_IV_2]], i64 1
-; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[INDVAR]], 1
; CHECK-NEXT: br i1 false, label [[INNER_3_LVER_CHECK:%.*]], label [[INNER_2]]
; CHECK: inner.3.lver.check:
-; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i64 [ [[INDVAR]], [[INNER_2]] ]
; CHECK-NEXT: [[LCSSA_PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2]], [[INNER_2]] ]
; CHECK-NEXT: [[GEP_6:%.*]] = getelementptr inbounds double, ptr [[PTR_PHI]], i64 1
; CHECK-NEXT: [[GEP_7:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_2]], i64 1
-; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDVAR_LCSSA]], 3
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 24
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[LCSSA_PTR_IV_1]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[LCSSA_PTR_IV_2]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 16
+; CHECK-NEXT: [[SCEVGEP:%.*]] = inttoptr i64 [[TMP1]] to ptr
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[GEP_7]], [[GEP_1]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[PTR_PHI]], [[SCEVGEP]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
@@ -90,10 +87,10 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK: inner.3:
; CHECK-NEXT: [[IV_2:%.*]] = phi i64 [ 0, [[INNER_3_PH]] ], [ [[IV_2_NEXT:%.*]], [[INNER_3]] ]
; CHECK-NEXT: [[GEP_8:%.*]] = getelementptr inbounds double, ptr [[GEP_6]], i64 [[IV_2]]
-; CHECK-NEXT: store double 0.000000e+00, ptr [[GEP_7]], align 8, !alias.scope !0, !noalias !3
-; CHECK-NEXT: store double 0.000000e+00, ptr [[GEP_8]], align 8, !alias.scope !3
+; CHECK-NEXT: store double 0.000000e+00, ptr [[GEP_7]], align 8, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
+; CHECK-NEXT: store double 0.000000e+00, ptr [[GEP_8]], align 8, !alias.scope [[META3]]
; CHECK-NEXT: [[GEP_9:%.*]] = getelementptr double, ptr [[PTR_PHI]], i64 [[IV_2]]
-; CHECK-NEXT: [[TMP18:%.*]] = load double, ptr [[GEP_9]], align 8, !alias.scope !3
+; CHECK-NEXT: [[TMP18:%.*]] = load double, ptr [[GEP_9]], align 8, !alias.scope [[META3]]
; CHECK-NEXT: [[IV_2_NEXT]] = add nuw nsw i64 [[IV_2]], 1
; CHECK-NEXT: [[C_2:%.*]] = icmp eq i64 [[IV_2]], 1
; CHECK-NEXT: br i1 [[C_2]], label [[OUTER_LATCH_LOOPEXIT3:%.*]], label [[INNER_3]]
>From 8fde62772afff00c23ae0ac2ecfae1cb90aca73e Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Tue, 22 Jul 2025 12:04:27 +0100
Subject: [PATCH 2/4] !fixup add comment
---
llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
index 97de12ae4c053..0038a8c8accf0 100644
--- a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
+++ b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
@@ -1240,6 +1240,10 @@ Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
ExitSCEV = SE.getPtrToIntExpr(ExitSCEV, STy);
else if (S->getType() != PN.getType())
continue;
+
+ // Check if we can re-use the existing PN, by adjusting it with an expanded
+ // offset, if the offset is simpler (for now just checks if it is
+ // AddRec-free).
const SCEV *Diff = SE.getMinusSCEV(S, ExitSCEV);
if (isa<SCEVCouldNotCompute>(Diff) ||
SCEVExprContains(Diff,
>From 535fb717a53261bdbc4c1297275470e429cb2bed Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Fri, 25 Jul 2025 10:02:23 +0100
Subject: [PATCH 3/4] !fixup restrict supported SCEV expressions more
---
.../Analysis/ScalarEvolutionPatternMatch.h | 6 +++
.../Utils/ScalarEvolutionExpander.cpp | 14 +++---
.../invalidate-laa-after-versioning.ll | 4 +-
.../reuse-lcssa-phi-scev-expansion.ll | 43 +++++++------------
.../invalidate-laa-after-versioning.ll | 4 +-
5 files changed, 32 insertions(+), 39 deletions(-)
diff --git a/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h b/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h
index 09e3945f5a8ff..bff77073ef4e1 100644
--- a/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h
+++ b/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h
@@ -160,6 +160,12 @@ m_scev_ZExt(const Op0_t &Op0) {
return m_scev_Unary<SCEVZeroExtendExpr>(Op0);
}
+template <typename Op0_t>
+inline SCEVUnaryExpr_match<SCEVPtrToIntExpr, Op0_t>
+m_scev_PtrToInt(const Op0_t &Op0) {
+ return SCEVUnaryExpr_match<SCEVPtrToIntExpr, Op0_t>(Op0);
+}
+
/// Match a binary SCEV.
template <typename SCEVTy, typename Op0_t, typename Op1_t>
struct SCEVBinaryExpr_match {
diff --git a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
index 0038a8c8accf0..205072d2260bc 100644
--- a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
+++ b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
@@ -18,6 +18,7 @@
#include "llvm/ADT/SmallSet.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/LoopInfo.h"
+#include "llvm/Analysis/ScalarEvolutionPatternMatch.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/IR/DataLayout.h"
@@ -42,6 +43,7 @@ cl::opt<unsigned> llvm::SCEVCheapExpansionBudget(
"controls the budget that is considered cheap (default = 4)"));
using namespace PatternMatch;
+using namespace SCEVPatternMatch;
PoisonFlags::PoisonFlags(const Instruction *I) {
NUW = false;
@@ -1242,18 +1244,18 @@ Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
continue;
// Check if we can re-use the existing PN, by adjusting it with an expanded
- // offset, if the offset is simpler (for now just checks if it is
- // AddRec-free).
+ // offset, if the offset is simpler.
const SCEV *Diff = SE.getMinusSCEV(S, ExitSCEV);
- if (isa<SCEVCouldNotCompute>(Diff) ||
- SCEVExprContains(Diff,
- [](const SCEV *S) { return isa<SCEVAddRecExpr>(S); }))
+ const SCEV *Op = Diff;
+ match(Diff, m_scev_Mul(m_scev_AllOnes(), m_SCEV(Op)));
+ match(Op, m_scev_PtrToInt(m_SCEV(Op)));
+ if (!isa<SCEVConstant, SCEVUnknown>(Op))
continue;
Value *DiffV = expand(Diff);
Value *BaseV = &PN;
if (DiffV->getType()->isIntegerTy() && PhiTy->isPointerTy())
- BaseV = Builder.CreatePtrToInt(BaseV, DiffV->getType());
+ return Builder.CreatePtrAdd(BaseV, DiffV);
return Builder.CreateAdd(BaseV, DiffV);
}
diff --git a/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll b/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
index 3ad262bb20910..037851f364e24 100644
--- a/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
+++ b/llvm/test/Transforms/LoopLoadElim/invalidate-laa-after-versioning.ll
@@ -66,9 +66,7 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK-NEXT: [[LCSSA_PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2]], [[INNER_2]] ]
; CHECK-NEXT: [[GEP_6:%.*]] = getelementptr inbounds double, ptr [[PTR_PHI]], i64 1
; CHECK-NEXT: [[GEP_7:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_2]], i64 1
-; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[LCSSA_PTR_IV_2]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 16
-; CHECK-NEXT: [[SCEVGEP3:%.*]] = inttoptr i64 [[TMP1]] to ptr
+; CHECK-NEXT: [[SCEVGEP3:%.*]] = getelementptr i8, ptr [[LCSSA_PTR_IV_2]], i64 16
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[GEP_7]], [[GEP_1]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[PTR_PHI]], [[SCEVGEP3]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
diff --git a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
index b9b21e986e422..e5dd534b298a3 100644
--- a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
+++ b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
@@ -114,22 +114,23 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-NEXT: br i1 [[EC_1]], label %[[LOOP_2_HEADER_PREHEADER:.*]], label %[[LOOP_1]]
; CHECK: [[LOOP_2_HEADER_PREHEADER]]:
; CHECK-NEXT: [[SEL_DST_LCSSA1:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1]] ]
+; CHECK-NEXT: [[PTR_IV_1_LCSSA1:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1]] ]
; CHECK-NEXT: [[SEL_DST_LCSSA:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
; CHECK-NEXT: [[SEL_DST_LCSSA12:%.*]] = ptrtoint ptr [[SEL_DST_LCSSA1]] to i64
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
+; CHECK-NEXT: [[TMP2:%.*]] = mul i64 [[SEL_DST_LCSSA12]], -1
+; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[PTR_IV_1_LCSSA]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[SEL_DST_LCSSA12]]
-; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP1]], 2
+; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 2
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA]], i64 1022
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 1022
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SEL_DST_LCSSA]], i64 1022
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA]], i64 [[INDEX]]
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 [[INDEX]]
; CHECK-NEXT: [[NEXT_GEP4:%.*]] = getelementptr i8, ptr [[SEL_DST_LCSSA]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[NEXT_GEP4]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i8>, ptr [[TMP4]], align 1
@@ -142,13 +143,13 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ 1023, %[[MIDDLE_BLOCK]] ], [ 1, %[[LOOP_2_HEADER_PREHEADER]] ], [ 1, %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL4:%.*]] = phi ptr [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA]], %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP1]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA1]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA1]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL6:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label %[[LOOP_2_HEADER:.*]]
; CHECK: [[LOOP_2_HEADER]]:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[DEC7:%.*]], %[[LOOP_2_LATCH:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL4]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL6]], %[[SCALAR_PH]] ]
; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i32 [[IV]], 1024
; CHECK-NEXT: br i1 [[EC_2]], label %[[EXIT:.*]], label %[[LOOP_2_LATCH]]
; CHECK: [[LOOP_2_LATCH]]:
@@ -209,10 +210,8 @@ define void @expand_diff_scev_unknown(ptr %dst, i1 %invar.c, i32 %step) mustprog
; CHECK-NEXT: br i1 [[INVAR_C]], label %[[LOOP_2_PREHEADER:.*]], label %[[LOOP_1]]
; CHECK: [[LOOP_2_PREHEADER]]:
; CHECK-NEXT: [[INDVAR_LCSSA1:%.*]] = phi i32 [ [[INDVAR]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i32 [ [[INDVAR]], %[[LOOP_1]] ]
; CHECK-NEXT: [[IV_1_LCSSA:%.*]] = phi i32 [ [[IV_1]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[STEP]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[INDVAR_LCSSA]], [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[IV_1_LCSSA]], [[STEP]]
; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP1]], i32 0)
; CHECK-NEXT: [[TMP2:%.*]] = mul i32 [[STEP]], -2
; CHECK-NEXT: [[TMP3:%.*]] = mul i32 [[INDVAR_LCSSA1]], -1
@@ -285,43 +284,33 @@ define void @expand_diff_neg_ptrtoint_expr(ptr %src, ptr %start) {
; CHECK-SAME: ptr [[SRC:%.*]], ptr [[START:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: [[SRC2:%.*]] = ptrtoint ptr [[SRC]] to i64
-; CHECK-NEXT: [[START1:%.*]] = ptrtoint ptr [[START]] to i64
; CHECK-NEXT: br label %[[LOOP_1:.*]]
; CHECK: [[LOOP_1]]:
-; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], %[[LOOP_1]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 1, %[[ENTRY]] ], [ [[IV_NEXT:%.*]], %[[LOOP_1]] ]
; CHECK-NEXT: [[PTR_IV_1:%.*]] = phi ptr [ [[START]], %[[ENTRY]] ], [ [[PTR_IV_1_NEXT:%.*]], %[[LOOP_1]] ]
; CHECK-NEXT: [[PTR_IV_1_NEXT]] = getelementptr i8, ptr [[PTR_IV_1]], i64 8
; CHECK-NEXT: call void @foo()
; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
; CHECK-NEXT: [[EC_1:%.*]] = icmp eq i64 [[IV_NEXT]], 32
-; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[INDVAR]], 1
; CHECK-NEXT: br i1 [[EC_1]], label %[[LOOP_2_PREHEADER:.*]], label %[[LOOP_1]]
; CHECK: [[LOOP_2_PREHEADER]]:
-; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i64 [ [[INDVAR]], %[[LOOP_1]] ]
; CHECK-NEXT: [[PTR_IV_1_NEXT_LCSSA:%.*]] = phi ptr [ [[PTR_IV_1_NEXT]], %[[LOOP_1]] ]
; CHECK-NEXT: br label %[[LOOP_2:.*]]
; CHECK: [[LOOP_2]]:
-; CHECK-NEXT: [[INDVAR3:%.*]] = phi i64 [ 0, %[[LOOP_2_PREHEADER]] ], [ [[INDVAR_NEXT4:%.*]], %[[LOOP_2]] ]
; CHECK-NEXT: [[IV_1:%.*]] = phi i64 [ [[IV_NEXT_1:%.*]], %[[LOOP_2]] ], [ 1, %[[LOOP_2_PREHEADER]] ]
; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2]] ], [ [[PTR_IV_1_NEXT_LCSSA]], %[[LOOP_2_PREHEADER]] ]
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: [[PTR_IV_2_NEXT]] = getelementptr i8, ptr [[PTR_IV_2]], i64 8
; CHECK-NEXT: [[IV_NEXT_1]] = add i64 [[IV_1]], 1
; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i64 [[IV_NEXT_1]], 32
-; CHECK-NEXT: [[INDVAR_NEXT4]] = add i64 [[INDVAR3]], 1
; CHECK-NEXT: br i1 [[EC_2]], label %[[LOOP_3_PREHEADER:.*]], label %[[LOOP_2]]
; CHECK: [[LOOP_3_PREHEADER]]:
-; CHECK-NEXT: [[INDVAR3_LCSSA:%.*]] = phi i64 [ [[INDVAR3]], %[[LOOP_2]] ]
; CHECK-NEXT: [[PTR_IV_2_NEXT_LCSSA:%.*]] = phi ptr [ [[PTR_IV_2_NEXT]], %[[LOOP_2]] ]
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[START1]], 16
-; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[SRC2]]
-; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[INDVAR_LCSSA]], 3
-; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[TMP2]], [[TMP1]]
-; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[INDVAR3_LCSSA]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = add i64 [[TMP4]], [[TMP3]]
+; CHECK-NEXT: [[TMP0:%.*]] = sub i64 0, [[SRC2]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[PTR_IV_2_NEXT_LCSSA]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP5:%.*]] = ptrtoint ptr [[TMP1]] to i64
; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP5]], 16
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
@@ -345,11 +334,11 @@ define void @expand_diff_neg_ptrtoint_expr(ptr %src, ptr %start) {
; CHECK-NEXT: br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ -1, %[[MIDDLE_BLOCK]] ], [ 1, %[[LOOP_3_PREHEADER]] ], [ 1, %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL6:%.*]] = phi ptr [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[LOOP_3_PREHEADER]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi ptr [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[LOOP_3_PREHEADER]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label %[[LOOP_3:.*]]
; CHECK: [[LOOP_3]]:
; CHECK-NEXT: [[IV_2:%.*]] = phi i64 [ [[IV_NEXT_2:%.*]], %[[LOOP_3]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_3]] ], [ [[BC_RESUME_VAL6]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_3]] ], [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ]
; CHECK-NEXT: [[TMP12:%.*]] = add i64 [[IV_2]], -1
; CHECK-NEXT: [[GEP_SRC:%.*]] = getelementptr double, ptr [[SRC]], i64 [[TMP12]]
; CHECK-NEXT: [[L:%.*]] = load i64, ptr [[GEP_SRC]], align 8
diff --git a/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll b/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
index 858864276c0a0..4148c3541ae7a 100644
--- a/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
+++ b/llvm/test/Transforms/LoopVersioning/invalidate-laa-after-versioning.ll
@@ -63,9 +63,7 @@ define void @test(ptr %arg, i64 %arg1) {
; CHECK-NEXT: [[LCSSA_PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2]], [[INNER_2]] ]
; CHECK-NEXT: [[GEP_6:%.*]] = getelementptr inbounds double, ptr [[PTR_PHI]], i64 1
; CHECK-NEXT: [[GEP_7:%.*]] = getelementptr inbounds double, ptr [[LCSSA_PTR_IV_2]], i64 1
-; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[LCSSA_PTR_IV_2]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], 16
-; CHECK-NEXT: [[SCEVGEP:%.*]] = inttoptr i64 [[TMP1]] to ptr
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[LCSSA_PTR_IV_2]], i64 16
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[GEP_7]], [[GEP_1]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[PTR_PHI]], [[SCEVGEP]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
>From 6854a9fe97cf6b5bc672ad7f06c516baea6fce0b Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Fri, 25 Jul 2025 11:44:57 +0100
Subject: [PATCH 4/4] !fixup add early exit, handle case with STy being int
---
.../Utils/ScalarEvolutionExpander.cpp | 10 +-
.../reuse-lcssa-phi-scev-expansion.ll | 98 +++++++++++++++++++
.../reuse-lcssa-phi-scev-expansion.ll | 31 +++---
3 files changed, 121 insertions(+), 18 deletions(-)
create mode 100644 llvm/test/Transforms/LoopIdiom/reuse-lcssa-phi-scev-expansion.ll
diff --git a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
index 205072d2260bc..d82a0d8f18915 100644
--- a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
+++ b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
@@ -1237,6 +1237,8 @@ Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
if (!SE.isSCEVable(PN.getType()))
continue;
auto *ExitSCEV = SE.getSCEV(&PN);
+ if (!isa<SCEVAddRecExpr>(ExitSCEV))
+ continue;
Type *PhiTy = PN.getType();
if (STy->isIntegerTy() && PhiTy->isPointerTy())
ExitSCEV = SE.getPtrToIntExpr(ExitSCEV, STy);
@@ -1252,10 +1254,14 @@ Value *SCEVExpander::tryToReuseLCSSAPhi(const SCEVAddRecExpr *S) {
if (!isa<SCEVConstant, SCEVUnknown>(Op))
continue;
+ assert(Diff->getType()->isIntegerTy() && "difference must be of integer type");
Value *DiffV = expand(Diff);
Value *BaseV = &PN;
- if (DiffV->getType()->isIntegerTy() && PhiTy->isPointerTy())
- return Builder.CreatePtrAdd(BaseV, DiffV);
+ if (PhiTy->isPointerTy()) {
+ if (STy->isPointerTy())
+ return Builder.CreatePtrAdd(BaseV, DiffV);
+ BaseV = Builder.CreatePtrToInt(BaseV, DiffV->getType());
+ }
return Builder.CreateAdd(BaseV, DiffV);
}
diff --git a/llvm/test/Transforms/LoopIdiom/reuse-lcssa-phi-scev-expansion.ll b/llvm/test/Transforms/LoopIdiom/reuse-lcssa-phi-scev-expansion.ll
new file mode 100644
index 0000000000000..357a7b6b7fd9a
--- /dev/null
+++ b/llvm/test/Transforms/LoopIdiom/reuse-lcssa-phi-scev-expansion.ll
@@ -0,0 +1,98 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -p loop-idiom -S %s | FileCheck %s
+
+declare void @foo()
+declare void @bar()
+
+define void @scev_expand_ptrtoint(i8 %x, ptr %start) {
+; CHECK-LABEL: define void @scev_expand_ptrtoint(
+; CHECK-SAME: i8 [[X:%.*]], ptr [[START:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[START1:%.*]] = ptrtoint ptr [[START]] to i64
+; CHECK-NEXT: br label %[[LOOP_1_HEADER:.*]]
+; CHECK: [[LOOP_1_HEADER]]:
+; CHECK-NEXT: [[PTR_IV_1:%.*]] = phi ptr [ [[START]], %[[ENTRY]] ], [ [[PTR_IV_1_NEXT:%.*]], %[[LOOP_1_LATCH:.*]] ]
+; CHECK-NEXT: [[C:%.*]] = icmp ule i8 [[X]], 1
+; CHECK-NEXT: br i1 [[C]], label %[[LOOP_1_LATCH]], label %[[MIDDLE:.*]]
+; CHECK: [[LOOP_1_LATCH]]:
+; CHECK-NEXT: [[PTR_IV_1_NEXT]] = getelementptr i8, ptr [[PTR_IV_1]], i64 1
+; CHECK-NEXT: br label %[[LOOP_1_HEADER]]
+; CHECK: [[MIDDLE]]:
+; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1_HEADER]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i8 [[X]], 0
+; CHECK-NEXT: [[CMP_EXT:%.*]] = zext i1 [[CMP]] to i64
+; CHECK-NEXT: [[GEP_START:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA]], i64 [[CMP_EXT]]
+; CHECK-NEXT: br label %[[LOOP_2_HEADER:.*]]
+; CHECK: [[LOOP_2_HEADER]]:
+; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], %[[LOOP_2_LATCH:.*]] ], [ 0, %[[MIDDLE]] ]
+; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[GEP_START]], %[[MIDDLE]] ], [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ]
+; CHECK-NEXT: switch i8 [[X]], label %[[LOOP_2_LATCH]] [
+; CHECK-NEXT: i8 1, label %[[LOOP_3_PREHEADER:.*]]
+; CHECK-NEXT: i8 4, label %[[LOOP_3_PREHEADER]]
+; CHECK-NEXT: ]
+; CHECK: [[LOOP_3_PREHEADER]]:
+; CHECK-NEXT: [[INDVAR_LCSSA:%.*]] = phi i64 [ [[INDVAR]], %[[LOOP_2_HEADER]] ], [ [[INDVAR]], %[[LOOP_2_HEADER]] ]
+; CHECK-NEXT: [[PTR_IV_2_LCSSA:%.*]] = phi ptr [ [[PTR_IV_2]], %[[LOOP_2_HEADER]] ], [ [[PTR_IV_2]], %[[LOOP_2_HEADER]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = sub i64 0, [[START1]]
+; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[PTR_IV_1_LCSSA]] to i64
+; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP1]], [[TMP0]]
+; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[TMP2]], 1
+; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[CMP_EXT]], [[TMP3]]
+; CHECK-NEXT: [[TMP5:%.*]] = add i64 [[INDVAR_LCSSA]], [[TMP4]]
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[START]], i64 [[TMP5]]
+; CHECK-NEXT: [[STRLEN:%.*]] = call i64 @strlen(ptr [[SCEVGEP]])
+; CHECK-NEXT: br label %[[LOOP_3:.*]]
+; CHECK: [[LOOP_2_LATCH]]:
+; CHECK-NEXT: [[PTR_IV_2_NEXT]] = getelementptr i8, ptr [[PTR_IV_2]], i64 1
+; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[INDVAR]], 1
+; CHECK-NEXT: br label %[[LOOP_2_HEADER]]
+; CHECK: [[LOOP_3]]:
+; CHECK-NEXT: [[IV_3:%.*]] = phi i64 [ [[IV_3_NEXT:%.*]], %[[LOOP_3]] ], [ 1, %[[LOOP_3_PREHEADER]] ]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_2_LCSSA]], i64 [[IV_3]]
+; CHECK-NEXT: [[TMP6:%.*]] = load i8, ptr [[GEP]], align 1
+; CHECK-NEXT: [[EC:%.*]] = icmp eq i8 [[TMP6]], 0
+; CHECK-NEXT: [[IV_3_NEXT]] = add i64 [[IV_3]], 1
+; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[LOOP_3]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %loop.1.header
+
+loop.1.header:
+ %ptr.iv.1 = phi ptr [ %start, %entry ], [ %ptr.iv.1.next, %loop.1.latch ]
+ %c = icmp ule i8 %x, 1
+ br i1 %c, label %loop.1.latch, label %middle
+
+loop.1.latch:
+ %ptr.iv.1.next = getelementptr i8, ptr %ptr.iv.1, i64 1
+ br label %loop.1.header
+
+middle:
+ %cmp = icmp eq i8 %x, 0
+ %cmp.ext = zext i1 %cmp to i64
+ %gep.start = getelementptr i8, ptr %ptr.iv.1, i64 %cmp.ext
+ br label %loop.2.header
+
+loop.2.header:
+ %ptr.iv.2 = phi ptr [ %gep.start, %middle ], [ %ptr.iv.2.next, %loop.2.latch ]
+ switch i8 %x, label %loop.2.latch [
+ i8 1, label %loop.3
+ i8 4, label %loop.3
+ ]
+
+loop.2.latch:
+ %ptr.iv.2.next = getelementptr i8, ptr %ptr.iv.2, i64 1
+ br label %loop.2.header
+
+loop.3:
+ %iv.3 = phi i64 [ 1, %loop.2.header ], [ 1, %loop.2.header ], [ %iv.3.next, %loop.3 ]
+ %gep = getelementptr i8, ptr %ptr.iv.2, i64 %iv.3
+ %1 = load i8, ptr %gep, align 1
+ %ec = icmp eq i8 %1, 0
+ %iv.3.next = add i64 %iv.3, 1
+ br i1 %ec, label %exit, label %loop.3
+
+exit:
+ ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
index e5dd534b298a3..4697b4abf5ea8 100644
--- a/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
+++ b/llvm/test/Transforms/LoopVectorize/reuse-lcssa-phi-scev-expansion.ll
@@ -114,23 +114,22 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-NEXT: br i1 [[EC_1]], label %[[LOOP_2_HEADER_PREHEADER:.*]], label %[[LOOP_1]]
; CHECK: [[LOOP_2_HEADER_PREHEADER]]:
; CHECK-NEXT: [[SEL_DST_LCSSA1:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
-; CHECK-NEXT: [[PTR_IV_1_LCSSA1:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1]] ]
+; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = phi ptr [ [[PTR_IV_1]], %[[LOOP_1]] ]
; CHECK-NEXT: [[SEL_DST_LCSSA:%.*]] = phi ptr [ [[SEL_DST]], %[[LOOP_1]] ]
; CHECK-NEXT: [[SEL_DST_LCSSA12:%.*]] = ptrtoint ptr [[SEL_DST_LCSSA1]] to i64
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
-; CHECK-NEXT: [[TMP2:%.*]] = mul i64 [[SEL_DST_LCSSA12]], -1
-; CHECK-NEXT: [[PTR_IV_1_LCSSA:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[PTR_IV_1_LCSSA]] to i64
-; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[SEL_DST_LCSSA12]]
+; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP1]], 2
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 1022
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA]], i64 1022
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SEL_DST_LCSSA]], i64 1022
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA1]], i64 [[INDEX]]
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_1_LCSSA]], i64 [[INDEX]]
; CHECK-NEXT: [[NEXT_GEP4:%.*]] = getelementptr i8, ptr [[SEL_DST_LCSSA]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[NEXT_GEP4]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i8>, ptr [[TMP4]], align 1
@@ -143,13 +142,13 @@ define void @runtime_checks_ptr_inductions(ptr %dst.1, ptr %dst.2, i1 %c) {
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ 1023, %[[MIDDLE_BLOCK]] ], [ 1, %[[LOOP_2_HEADER_PREHEADER]] ], [ 1, %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP1]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA1]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA1]], %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL6:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL4:%.*]] = phi ptr [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_1_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[PTR_IV_1_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL5:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[SEL_DST_LCSSA]], %[[LOOP_2_HEADER_PREHEADER]] ], [ [[SEL_DST_LCSSA]], %[[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label %[[LOOP_2_HEADER:.*]]
; CHECK: [[LOOP_2_HEADER]]:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[DEC7:%.*]], %[[LOOP_2_LATCH:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL6]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_2:%.*]] = phi ptr [ [[PTR_IV_2_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL4]], %[[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR_IV_3:%.*]] = phi ptr [ [[PTR_IV_3_NEXT:%.*]], %[[LOOP_2_LATCH]] ], [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ]
; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i32 [[IV]], 1024
; CHECK-NEXT: br i1 [[EC_2]], label %[[EXIT:.*]], label %[[LOOP_2_LATCH]]
; CHECK: [[LOOP_2_LATCH]]:
@@ -305,22 +304,22 @@ define void @expand_diff_neg_ptrtoint_expr(ptr %src, ptr %start) {
; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i64 [[IV_NEXT_1]], 32
; CHECK-NEXT: br i1 [[EC_2]], label %[[LOOP_3_PREHEADER:.*]], label %[[LOOP_2]]
; CHECK: [[LOOP_3_PREHEADER]]:
-; CHECK-NEXT: [[PTR_IV_2_NEXT_LCSSA:%.*]] = phi ptr [ [[PTR_IV_2_NEXT]], %[[LOOP_2]] ]
+; CHECK-NEXT: [[TMP1:%.*]] = phi ptr [ [[PTR_IV_2_NEXT]], %[[LOOP_2]] ]
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK: [[VECTOR_MEMCHECK]]:
; CHECK-NEXT: [[TMP0:%.*]] = sub i64 0, [[SRC2]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[PTR_IV_2_NEXT_LCSSA]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP5:%.*]] = ptrtoint ptr [[TMP1]] to i64
-; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP5]], 16
+; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP5]], [[TMP0]]
+; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP2]], 16
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[PTR_IV_2_NEXT_LCSSA]], i64 -16
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 -16
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
; CHECK-NEXT: [[OFFSET_IDX5:%.*]] = mul i64 [[INDEX]], 8
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PTR_IV_2_NEXT_LCSSA]], i64 [[OFFSET_IDX5]]
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[OFFSET_IDX5]]
; CHECK-NEXT: [[TMP7:%.*]] = add i64 [[OFFSET_IDX]], -1
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[SRC]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i64, ptr [[TMP8]], i32 0
@@ -334,7 +333,7 @@ define void @expand_diff_neg_ptrtoint_expr(ptr %src, ptr %start) {
; CHECK-NEXT: br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ -1, %[[MIDDLE_BLOCK]] ], [ 1, %[[LOOP_3_PREHEADER]] ], [ 1, %[[VECTOR_MEMCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi ptr [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[LOOP_3_PREHEADER]] ], [ [[PTR_IV_2_NEXT_LCSSA]], %[[VECTOR_MEMCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi ptr [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ [[TMP1]], %[[LOOP_3_PREHEADER]] ], [ [[TMP1]], %[[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label %[[LOOP_3:.*]]
; CHECK: [[LOOP_3]]:
; CHECK-NEXT: [[IV_2:%.*]] = phi i64 [ [[IV_NEXT_2:%.*]], %[[LOOP_3]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
More information about the llvm-commits
mailing list