Skip to content

Commit

Permalink
Reland "[InstCombine] Recognize ((x * y) s/ x) !=/== y as an signed…
Browse files Browse the repository at this point in the history 
… multiplication overflow check (PR48769)"

This reverts commit 91f7a4f,
relanding commit 13ec913.

The original commit was reverted because of (essentially)
https://bugs.llvm.org/show_bug.cgi?id=35922
which has now been addressed by d0eeb64.
  • Loading branch information
@LebedevRI
LebedevRI committed Sep 7, 2021
1 parent 0a5ebc6 commit 35fa7b8
Show file tree
Hide file tree
Showing 4 changed files with 71 additions and 66 deletions.
39 changes: 24 additions & 15 deletions llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
View file
Expand Up @@ -3829,19 +3829,22 @@ foldShiftIntoShiftInAnotherHandOfAndInICmp(ICmpInst &I, const SimplifyQuery SQ,

/// Fold
/// (-1 u/ x) u< y
/// ((x * y) u/ x) != y
/// ((x * y) ?/ x) != y
/// to
/// @llvm.umul.with.overflow(x, y) plus extraction of overflow bit
/// @llvm.?mul.with.overflow(x, y) plus extraction of overflow bit
/// Note that the comparison is commutative, while inverted (u>=, ==) predicate
/// will mean that we are looking for the opposite answer.
Value *InstCombinerImpl::foldUnsignedMultiplicationOverflowCheck(ICmpInst &I) {
Value *InstCombinerImpl::foldMultiplicationOverflowCheck(ICmpInst &I) {
ICmpInst::Predicate Pred;
Value *X, *Y;
Instruction *Mul;
Instruction *Div;
bool NeedNegation;
// Look for: (-1 u/ x) u</u>= y
if (!I.isEquality() &&
match(&I, m_c_ICmp(Pred, m_OneUse(m_UDiv(m_AllOnes(), m_Value(X))),
match(&I, m_c_ICmp(Pred,
m_CombineAnd(m_OneUse(m_UDiv(m_AllOnes(), m_Value(X))),
m_Instruction(Div)),
m_Value(Y)))) {
Mul = nullptr;

Expand All @@ -3856,13 +3859,16 @@ Value *InstCombinerImpl::foldUnsignedMultiplicationOverflowCheck(ICmpInst &I) {
default:
return nullptr; // Wrong predicate.
}
} else // Look for: ((x * y) u/ x) !=/== y
} else // Look for: ((x * y) / x) !=/== y
if (I.isEquality() &&
match(&I, m_c_ICmp(Pred, m_Value(Y),
m_OneUse(m_UDiv(m_CombineAnd(m_c_Mul(m_Deferred(Y),
match(&I,
m_c_ICmp(Pred, m_Value(Y),
m_CombineAnd(
m_OneUse(m_IDiv(m_CombineAnd(m_c_Mul(m_Deferred(Y),
m_Value(X)),
m_Instruction(Mul)),
m_Deferred(X)))))) {
m_Deferred(X))),
m_Instruction(Div))))) {
NeedNegation = Pred == ICmpInst::Predicate::ICMP_EQ;
} else
return nullptr;
Expand All @@ -3874,19 +3880,22 @@ Value *InstCombinerImpl::foldUnsignedMultiplicationOverflowCheck(ICmpInst &I) {
if (MulHadOtherUses)
Builder.SetInsertPoint(Mul);

Function *F = Intrinsic::getDeclaration(
I.getModule(), Intrinsic::umul_with_overflow, X->getType());
CallInst *Call = Builder.CreateCall(F, {X, Y}, "umul");
Function *F = Intrinsic::getDeclaration(I.getModule(),
Div->getOpcode() == Instruction::UDiv
? Intrinsic::umul_with_overflow
: Intrinsic::smul_with_overflow,
X->getType());
CallInst *Call = Builder.CreateCall(F, {X, Y}, "mul");

// If the multiplication was used elsewhere, to ensure that we don't leave
// "duplicate" instructions, replace uses of that original multiplication
// with the multiplication result from the with.overflow intrinsic.
if (MulHadOtherUses)
replaceInstUsesWith(*Mul, Builder.CreateExtractValue(Call, 0, "umul.val"));
replaceInstUsesWith(*Mul, Builder.CreateExtractValue(Call, 0, "mul.val"));

Value *Res = Builder.CreateExtractValue(Call, 1, "umul.ov");
Value *Res = Builder.CreateExtractValue(Call, 1, "mul.ov");
if (NeedNegation) // This technically increases instruction count.
Res = Builder.CreateNot(Res, "umul.not.ov");
Res = Builder.CreateNot(Res, "mul.not.ov");

// If we replaced the mul, erase it. Do this after all uses of Builder,
// as the mul is used as insertion point.
Expand Down Expand Up @@ -4283,7 +4292,7 @@ Instruction *InstCombinerImpl::foldICmpBinOp(ICmpInst &I,
}
}

if (Value *V = foldUnsignedMultiplicationOverflowCheck(I))
if (Value *V = foldMultiplicationOverflowCheck(I))
return replaceInstUsesWith(I, V);

if (Value *V = foldICmpWithLowBitMaskedVal(I, Builder))
Expand Down
2 changes: 1 addition & 1 deletion llvm/lib/Transforms/InstCombine/InstCombineInternal.h
View file
Expand Up @@ -661,7 +661,7 @@ class LLVM_LIBRARY_VISIBILITY InstCombinerImpl final
Instruction *foldSignBitTest(ICmpInst &I);
Instruction *foldICmpWithZero(ICmpInst &Cmp);

Value *foldUnsignedMultiplicationOverflowCheck(ICmpInst &Cmp);
Value *foldMultiplicationOverflowCheck(ICmpInst &Cmp);

Instruction *foldICmpSelectConstant(ICmpInst &Cmp, SelectInst *Select,
ConstantInt *C);
Expand Down
51 changes: 26 additions & 25 deletions llvm/test/Transforms/InstCombine/signed-mul-lack-of-overflow-check-via-mul-sdiv.ll
View file
Expand Up @@ -8,10 +8,10 @@

define i1 @t0_basic(i8 %x, i8 %y) {
; CHECK-LABEL: @t0_basic(
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y:%.*]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor i1 [[MUL_OV]], true
; CHECK-NEXT: ret i1 [[MUL_NOT_OV]]
;
%t0 = mul i8 %x, %y
%t1 = sdiv i8 %t0, %x
Expand All @@ -21,10 +21,10 @@ define i1 @t0_basic(i8 %x, i8 %y) {

define <2 x i1> @t1_vec(<2 x i8> %x, <2 x i8> %y) {
; CHECK-LABEL: @t1_vec(
; CHECK-NEXT: [[T0:%.*]] = mul <2 x i8> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv <2 x i8> [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[T1]], [[Y]]
; CHECK-NEXT: ret <2 x i1> [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { <2 x i8>, <2 x i1> } @llvm.smul.with.overflow.v2i8(<2 x i8> [[X:%.*]], <2 x i8> [[Y:%.*]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { <2 x i8>, <2 x i1> } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor <2 x i1> [[MUL_OV]], <i1 true, i1 true>
; CHECK-NEXT: ret <2 x i1> [[MUL_NOT_OV]]
;
%t0 = mul <2 x i8> %x, %y
%t1 = sdiv <2 x i8> %t0, %x
Expand All @@ -37,10 +37,10 @@ declare i8 @gen8()
define i1 @t2_commutative(i8 %x) {
; CHECK-LABEL: @t2_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor i1 [[MUL_OV]], true
; CHECK-NEXT: ret i1 [[MUL_NOT_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -52,10 +52,10 @@ define i1 @t2_commutative(i8 %x) {
define i1 @t3_commutative(i8 %x) {
; CHECK-LABEL: @t3_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor i1 [[MUL_OV]], true
; CHECK-NEXT: ret i1 [[MUL_NOT_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -67,10 +67,10 @@ define i1 @t3_commutative(i8 %x) {
define i1 @t4_commutative(i8 %x) {
; CHECK-LABEL: @t4_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq i8 [[Y]], [[T1]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor i1 [[MUL_OV]], true
; CHECK-NEXT: ret i1 [[MUL_NOT_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -85,11 +85,12 @@ declare void @use8(i8)

define i1 @t5_extrause0(i8 %x, i8 %y) {
; CHECK-LABEL: @t5_extrause0(
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: call void @use8(i8 [[T0]])
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp eq i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y:%.*]])
; CHECK-NEXT: [[MUL_VAL:%.*]] = extractvalue { i8, i1 } [[MUL]], 0
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: [[MUL_NOT_OV:%.*]] = xor i1 [[MUL_OV]], true
; CHECK-NEXT: call void @use8(i8 [[MUL_VAL]])
; CHECK-NEXT: ret i1 [[MUL_NOT_OV]]
;
%t0 = mul i8 %x, %y
call void @use8(i8 %t0)
Expand Down
45 changes: 20 additions & 25 deletions llvm/test/Transforms/InstCombine/signed-mul-overflow-check-via-mul-sdiv.ll
View file
Expand Up @@ -8,10 +8,9 @@

define i1 @t0_basic(i8 %x, i8 %y) {
; CHECK-LABEL: @t0_basic(
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y:%.*]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: ret i1 [[MUL_OV]]
;
%t0 = mul i8 %x, %y
%t1 = sdiv i8 %t0, %x
Expand All @@ -21,10 +20,9 @@ define i1 @t0_basic(i8 %x, i8 %y) {

define <2 x i1> @t1_vec(<2 x i8> %x, <2 x i8> %y) {
; CHECK-LABEL: @t1_vec(
; CHECK-NEXT: [[T0:%.*]] = mul <2 x i8> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv <2 x i8> [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne <2 x i8> [[T1]], [[Y]]
; CHECK-NEXT: ret <2 x i1> [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { <2 x i8>, <2 x i1> } @llvm.smul.with.overflow.v2i8(<2 x i8> [[X:%.*]], <2 x i8> [[Y:%.*]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { <2 x i8>, <2 x i1> } [[MUL]], 1
; CHECK-NEXT: ret <2 x i1> [[MUL_OV]]
;
%t0 = mul <2 x i8> %x, %y
%t1 = sdiv <2 x i8> %t0, %x
Expand All @@ -37,10 +35,9 @@ declare i8 @gen8()
define i1 @t2_commutative(i8 %x) {
; CHECK-LABEL: @t2_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: ret i1 [[MUL_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -52,10 +49,9 @@ define i1 @t2_commutative(i8 %x) {
define i1 @t3_commutative(i8 %x) {
; CHECK-LABEL: @t3_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: ret i1 [[MUL_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -67,10 +63,9 @@ define i1 @t3_commutative(i8 %x) {
define i1 @t4_commutative(i8 %x) {
; CHECK-LABEL: @t4_commutative(
; CHECK-NEXT: [[Y:%.*]] = call i8 @gen8()
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[Y]], [[X:%.*]]
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne i8 [[Y]], [[T1]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y]])
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: ret i1 [[MUL_OV]]
;
%y = call i8 @gen8()
%t0 = mul i8 %y, %x ; swapped
Expand All @@ -85,11 +80,11 @@ declare void @use8(i8)

define i1 @t5_extrause0(i8 %x, i8 %y) {
; CHECK-LABEL: @t5_extrause0(
; CHECK-NEXT: [[T0:%.*]] = mul i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: call void @use8(i8 [[T0]])
; CHECK-NEXT: [[T1:%.*]] = sdiv i8 [[T0]], [[X]]
; CHECK-NEXT: [[R:%.*]] = icmp ne i8 [[T1]], [[Y]]
; CHECK-NEXT: ret i1 [[R]]
; CHECK-NEXT: [[MUL:%.*]] = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 [[X:%.*]], i8 [[Y:%.*]])
; CHECK-NEXT: [[MUL_VAL:%.*]] = extractvalue { i8, i1 } [[MUL]], 0
; CHECK-NEXT: [[MUL_OV:%.*]] = extractvalue { i8, i1 } [[MUL]], 1
; CHECK-NEXT: call void @use8(i8 [[MUL_VAL]])
; CHECK-NEXT: ret i1 [[MUL_OV]]
;
%t0 = mul i8 %x, %y
call void @use8(i8 %t0)
Expand Down

0 comments on commit 35fa7b8

Please sign in to comment.