Skip to content

[CIR] Upstream Unary Inc/Dec for ComplexType #149162

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jul 17, 2025
Merged

[CIR] Upstream Unary Inc/Dec for ComplexType #149162

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f4ceabd
[CIR] Upstream Unary Inc/Dec for ComplexType
AmrDeveloper Jul 16, 2025
c034fc1
Update test to be more readable
AmrDeveloper Jul 17, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
49 changes: 49 additions & 0 deletions clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,26 @@ class ComplexExprEmitter : public StmtVisitor<ComplexExprEmitter, mlir::Value> {
mlir::Value VisitParenExpr(ParenExpr *e);
mlir::Value
VisitSubstNonTypeTemplateParmExpr(SubstNonTypeTemplateParmExpr *e);

mlir::Value VisitPrePostIncDec(const UnaryOperator *e, bool isInc,
bool isPre);

mlir::Value VisitUnaryPostDec(const UnaryOperator *e) {
Copy link
Contributor

@xlauko xlauko Jul 18, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I know this is already merged, but it would be nicer to pass cir::UnaryOpKind as second argument instead of ambiguous bool. @AmrDeveloper can you fix this in additional PR please?

Copy link
Member Author

@AmrDeveloper AmrDeveloper Jul 18, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure, I will create another PR for it, Thanks

Copy link
Member Author

@AmrDeveloper AmrDeveloper Jul 18, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I implemented it locally, but I was thinking that there are no checks on unary op, and anyone can call the function with cir::UnaryOpKind::Not, Minus or Plus and it will generate CIR code that will crash only when we lower to LLVM 🤔, do you think this is the best option or maybe we should rename the booleans in a better way @xlauko

Copy link
Contributor

@xlauko xlauko Jul 18, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can assert to expect inc/dec?

Copy link
Member Author

@AmrDeveloper AmrDeveloper Jul 18, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can assert to expect inc/dec?

Mmmmm, Yes, I will create a PR now

return VisitPrePostIncDec(e, false, false);
}

mlir::Value VisitUnaryPostInc(const UnaryOperator *e) {
return VisitPrePostIncDec(e, true, false);
}

mlir::Value VisitUnaryPreDec(const UnaryOperator *e) {
return VisitPrePostIncDec(e, false, true);
}

mlir::Value VisitUnaryPreInc(const UnaryOperator *e) {
return VisitPrePostIncDec(e, true, true);
}

mlir::Value VisitUnaryDeref(const Expr *e);
mlir::Value VisitUnaryNot(const UnaryOperator *e);

Expand Down Expand Up @@ -335,6 +355,12 @@ mlir::Value ComplexExprEmitter::VisitSubstNonTypeTemplateParmExpr(
return Visit(e->getReplacement());
}

mlir::Value ComplexExprEmitter::VisitPrePostIncDec(const UnaryOperator *e,
bool isInc, bool isPre) {
LValue lv = cgf.emitLValue(e->getSubExpr());
return cgf.emitComplexPrePostIncDec(e, lv, isInc, isPre);
}

mlir::Value ComplexExprEmitter::VisitUnaryDeref(const Expr *e) {
return emitLoadOfLValue(e);
}
Expand Down Expand Up @@ -423,6 +449,29 @@ mlir::Value CIRGenFunction::emitComplexExpr(const Expr *e) {
return ComplexExprEmitter(*this).Visit(const_cast<Expr *>(e));
}

mlir::Value CIRGenFunction::emitComplexPrePostIncDec(const UnaryOperator *e,
LValue lv, bool isInc,
bool isPre) {
mlir::Value inVal = emitLoadOfComplex(lv, e->getExprLoc());
mlir::Location loc = getLoc(e->getExprLoc());
auto opKind = isInc ? cir::UnaryOpKind::Inc : cir::UnaryOpKind::Dec;
mlir::Value incVal = builder.createUnaryOp(loc, opKind, inVal);

// Store the updated result through the lvalue.
emitStoreOfComplex(loc, incVal, lv, /*isInit=*/false);

if (getLangOpts().OpenMP)
cgm.errorNYI(loc, "emitComplexPrePostIncDec OpenMP");

// If this is a postinc, return the value read from memory, otherwise use the
// updated value.
return isPre ? incVal : inVal;
}

mlir::Value CIRGenFunction::emitLoadOfComplex(LValue src, SourceLocation loc) {
return ComplexExprEmitter(*this).emitLoadOfLValue(src, loc);
}

void CIRGenFunction::emitStoreOfComplex(mlir::Location loc, mlir::Value v,
LValue dest, bool isInit) {
ComplexExprEmitter(*this).emitStoreOfComplex(loc, v, dest, isInit);
Expand Down
6 changes: 6 additions & 0 deletions clang/lib/CIR/CodeGen/CIRGenFunction.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -930,6 +930,9 @@ class CIRGenFunction : public CIRGenTypeCache {
/// returning the result.
mlir::Value emitComplexExpr(const Expr *e);

mlir::Value emitComplexPrePostIncDec(const UnaryOperator *e, LValue lv,
bool isInc, bool isPre);

LValue emitComplexAssignmentLValue(const BinaryOperator *e);

void emitCompoundStmt(const clang::CompoundStmt &s);
Expand Down Expand Up @@ -980,6 +983,9 @@ class CIRGenFunction : public CIRGenTypeCache {

RValue emitLoadOfBitfieldLValue(LValue lv, SourceLocation loc);

/// Load a complex number from the specified l-value.
mlir::Value emitLoadOfComplex(LValue src, SourceLocation loc);

/// Given an expression that represents a value lvalue, this method emits
/// the address of the lvalue, then loads the result as an rvalue,
/// returning the rvalue.
Expand Down
3 changes: 2 additions & 1 deletion clang/lib/CIR/Dialect/Transforms/LoweringPrepare.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,8 @@ void LoweringPreparePass::lowerUnaryOp(cir::UnaryOp op) {
switch (opKind) {
case cir::UnaryOpKind::Inc:
case cir::UnaryOpKind::Dec:
llvm_unreachable("Complex unary Inc/Dec NYI");
resultReal = builder.createUnaryOp(loc, opKind, operandReal);
resultImag = operandImag;
break;

case cir::UnaryOpKind::Plus:
Expand Down
198 changes: 197 additions & 1 deletion clang/test/CIR/CodeGen/complex-unary.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,8 +83,204 @@ void foo2() {
// OGCG: %[[A_REAL:.*]] = load float, ptr %[[A_REAL_PTR]], align 4
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: %[[A_IMAG:.*]] = load float, ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[A_IMAG_MINUS:.*]] = fneg float %[[A_IMAG]]
// OGCG: %[[A_IMAG_MINUS:.*]] = fneg float %[[A_IMAG]]
// OGCG: %[[RESULT_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 0
// OGCG: %[[RESULT_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 1
// OGCG: store float %[[A_REAL]], ptr %[[RESULT_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG_MINUS]], ptr %[[RESULT_IMAG_PTR]], align 4

void foo3() {
float _Complex a;
float _Complex b = a++;
}

// CIR-BEFORE: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-BEFORE: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-BEFORE: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-BEFORE: %[[COMPLEX_INC:.*]] = cir.unary(inc, %[[TMP]]) : !cir.complex<!cir.float>, !cir.complex<!cir.float>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_INC]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-BEFORE: cir.store{{.*}} %[[TMP]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// CIR-AFTER: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-AFTER: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-AFTER: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-AFTER: %[[REAL:.*]] = cir.complex.real %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[IMAG:.*]] = cir.complex.imag %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[REAL_INC:.*]] = cir.unary(inc, %[[REAL]]) : !cir.float, !cir.float
// CIR-AFTER: %[[NEW_COMPLEX:.*]] = cir.complex.create %[[REAL_INC]], %[[IMAG]] : !cir.float -> !cir.complex<!cir.float>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-AFTER: cir.store{{.*}} %[[TMP]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// LLVM: %[[COMPLEX:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[RESULT:.*]] = alloca { float, float }, i64 1, align 4
Copy link
Contributor

@andykaylor andykaylor Jul 16, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// LLVM: %[[COMPLEX:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[RESULT:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[A_ADDR:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[B_ADDR:.*]] = alloca { float, float }, i64 1, align 4

This, along with other changes this imples, should make the checks more readable. And, of course, I'm suggesting this change throughout the new tests.

// LLVM: %[[TMP:.*]] = load { float, float }, ptr %[[COMPLEX]], align 4
// LLVM: %[[REAL:.*]] = extractvalue { float, float } %[[TMP]], 0
// LLVM: %[[IMAG:.*]] = extractvalue { float, float } %[[TMP]], 1
// LLVM: %[[REAL_INC:.*]] = fadd float 1.000000e+00, %[[REAL]]
// LLVM: %[[RESULT_TMP:.*]] = insertvalue { float, float } {{.*}}, float %[[REAL_INC]], 0
// LLVM: %[[RESULT_VAL:.*]] = insertvalue { float, float } %[[RESULT_TMP]], float %[[IMAG]], 1
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[COMPLEX]], align 4
// LLVM: store { float, float } %[[TMP]], ptr %[[RESULT]], align 4

// OGCG: %[[COMPLEX:.*]] = alloca { float, float }, align 4
// OGCG: %[[RESULT:.*]] = alloca { float, float }, align 4
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_REAL:.*]] = load float, ptr %[[A_REAL_PTR]], align 4
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: %[[A_IMAG:.*]] = load float, ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[A_REAL_INC:.*]] = fadd float %[[A_REAL]], 1.000000e+00
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: store float %[[A_REAL_INC]], ptr %[[A_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[RESULT_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 0
// OGCG: %[[RESULT_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 1
// OGCG: store float %[[A_REAL]], ptr %[[RESULT_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[RESULT_IMAG_PTR]], align 4

void foo4() {
float _Complex a;
float _Complex b = ++a;
}

// CIR-BEFORE: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-BEFORE: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-BEFORE: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-BEFORE: %[[COMPLEX_INC:.*]] = cir.unary(inc, %[[TMP]]) : !cir.complex<!cir.float>, !cir.complex<!cir.float>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_INC]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_INC]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// CIR-AFTER: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-AFTER: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-AFTER: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-AFTER: %[[REAL:.*]] = cir.complex.real %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[IMAG:.*]] = cir.complex.imag %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[REAL_INC:.*]] = cir.unary(inc, %[[REAL]]) : !cir.float, !cir.float
// CIR-AFTER: %[[NEW_COMPLEX:.*]] = cir.complex.create %[[REAL_INC]], %[[IMAG]] : !cir.float -> !cir.complex<!cir.float>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// LLVM: %[[COMPLEX:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[RESULT:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[TMP:.*]] = load { float, float }, ptr %[[COMPLEX]], align 4
// LLVM: %[[REAL:.*]] = extractvalue { float, float } %[[TMP]], 0
// LLVM: %[[IMAG:.*]] = extractvalue { float, float } %[[TMP]], 1
// LLVM: %[[REAL_INC:.*]] = fadd float 1.000000e+00, %[[REAL]]
// LLVM: %[[RESULT_TMP:.*]] = insertvalue { float, float } {{.*}}, float %[[REAL_INC]], 0
// LLVM: %[[RESULT_VAL:.*]] = insertvalue { float, float } %[[RESULT_TMP]], float %[[IMAG]], 1
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[COMPLEX]], align 4
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[RESULT]], align 4

// OGCG: %[[COMPLEX:.*]] = alloca { float, float }, align 4
// OGCG: %[[RESULT:.*]] = alloca { float, float }, align 4
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_REAL:.*]] = load float, ptr %[[A_REAL_PTR]], align 4
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: %[[A_IMAG:.*]] = load float, ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[A_REAL_INC:.*]] = fadd float %[[A_REAL]], 1.000000e+00
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: store float %[[A_REAL_INC]], ptr %[[A_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[RESULT_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 0
// OGCG: %[[RESULT_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 1
// OGCG: store float %[[A_REAL_INC]], ptr %[[RESULT_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[RESULT_IMAG_PTR]], align 4

void foo5() {
float _Complex a;
float _Complex b = a--;
}

// CIR-BEFORE: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-BEFORE: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-BEFORE: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-BEFORE: %[[COMPLEX_DEC:.*]] = cir.unary(dec, %[[TMP]]) : !cir.complex<!cir.float>, !cir.complex<!cir.float>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_DEC]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-BEFORE: cir.store{{.*}} %[[TMP]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// CIR-AFTER: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-AFTER: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-AFTER: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-AFTER: %[[REAL:.*]] = cir.complex.real %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[IMAG:.*]] = cir.complex.imag %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[REAL_DEC:.*]] = cir.unary(dec, %[[REAL]]) : !cir.float, !cir.float
// CIR-AFTER: %[[NEW_COMPLEX:.*]] = cir.complex.create %[[REAL_DEC]], %[[IMAG]] : !cir.float -> !cir.complex<!cir.float>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-AFTER: cir.store{{.*}} %[[TMP]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// LLVM: %[[COMPLEX:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[RESULT:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[TMP:.*]] = load { float, float }, ptr %[[COMPLEX]], align 4
// LLVM: %[[REAL:.*]] = extractvalue { float, float } %[[TMP]], 0
// LLVM: %[[IMAG:.*]] = extractvalue { float, float } %[[TMP]], 1
// LLVM: %[[REAL_DEC:.*]] = fadd float -1.000000e+00, %[[REAL]]
// LLVM: %[[RESULT_TMP:.*]] = insertvalue { float, float } {{.*}}, float %[[REAL_DEC]], 0
// LLVM: %[[RESULT_VAL:.*]] = insertvalue { float, float } %[[RESULT_TMP]], float %[[IMAG]], 1
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[COMPLEX]], align 4
// LLVM: store { float, float } %[[TMP]], ptr %[[RESULT]], align 4

// OGCG: %[[COMPLEX:.*]] = alloca { float, float }, align 4
// OGCG: %[[RESULT:.*]] = alloca { float, float }, align 4
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_REAL:.*]] = load float, ptr %[[A_REAL_PTR]], align 4
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: %[[A_IMAG:.*]] = load float, ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[A_REAL_DEC:.*]] = fadd float %[[A_REAL]], -1.000000e+00
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: store float %[[A_REAL_DEC]], ptr %[[A_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[RESULT_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 0
// OGCG: %[[RESULT_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 1
// OGCG: store float %[[A_REAL]], ptr %[[RESULT_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[RESULT_IMAG_PTR]], align 4

void foo6() {
float _Complex a;
float _Complex b = --a;
}

// CIR-BEFORE: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-BEFORE: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-BEFORE: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-BEFORE: %[[COMPLEX_DEC:.*]] = cir.unary(dec, %[[TMP]]) : !cir.complex<!cir.float>, !cir.complex<!cir.float>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_DEC]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-BEFORE: cir.store{{.*}} %[[COMPLEX_DEC]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// CIR-AFTER: %[[COMPLEX:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["a"]
// CIR-AFTER: %[[RESULT:.*]] = cir.alloca !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>, ["b", init]
// CIR-AFTER: %[[TMP:.*]] = cir.load{{.*}} %[[COMPLEX]] : !cir.ptr<!cir.complex<!cir.float>>, !cir.complex<!cir.float>
// CIR-AFTER: %[[REAL:.*]] = cir.complex.real %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[IMAG:.*]] = cir.complex.imag %[[TMP]] : !cir.complex<!cir.float> -> !cir.float
// CIR-AFTER: %[[REAL_DEC:.*]] = cir.unary(dec, %[[REAL]]) : !cir.float, !cir.float
// CIR-AFTER: %[[NEW_COMPLEX:.*]] = cir.complex.create %[[REAL_DEC]], %[[IMAG]] : !cir.float -> !cir.complex<!cir.float>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[COMPLEX]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>
// CIR-AFTER: cir.store{{.*}} %[[NEW_COMPLEX]], %[[RESULT]] : !cir.complex<!cir.float>, !cir.ptr<!cir.complex<!cir.float>>

// LLVM: %[[COMPLEX:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[RESULT:.*]] = alloca { float, float }, i64 1, align 4
// LLVM: %[[TMP:.*]] = load { float, float }, ptr %[[COMPLEX]], align 4
// LLVM: %[[REAL:.*]] = extractvalue { float, float } %[[TMP]], 0
// LLVM: %[[IMAG:.*]] = extractvalue { float, float } %[[TMP]], 1
// LLVM: %[[REAL_DEC:.*]] = fadd float -1.000000e+00, %[[REAL]]
// LLVM: %[[RESULT_TMP:.*]] = insertvalue { float, float } {{.*}}, float %[[REAL_DEC]], 0
// LLVM: %[[RESULT_VAL:.*]] = insertvalue { float, float } %[[RESULT_TMP]], float %[[IMAG]], 1
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[COMPLEX]], align 4
// LLVM: store { float, float } %[[RESULT_VAL]], ptr %[[RESULT]], align 4

// OGCG: %[[COMPLEX:.*]] = alloca { float, float }, align 4
// OGCG: %[[RESULT:.*]] = alloca { float, float }, align 4
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_REAL:.*]] = load float, ptr %[[A_REAL_PTR]], align 4
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: %[[A_IMAG:.*]] = load float, ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[A_REAL_DEC:.*]] = fadd float %[[A_REAL]], -1.000000e+00
// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 0
// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[COMPLEX]], i32 0, i32 1
// OGCG: store float %[[A_REAL_DEC]], ptr %[[A_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[A_IMAG_PTR]], align 4
// OGCG: %[[RESULT_REAL_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 0
// OGCG: %[[RESULT_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[RESULT]], i32 0, i32 1
// OGCG: store float %[[A_REAL_DEC]], ptr %[[RESULT_REAL_PTR]], align 4
// OGCG: store float %[[A_IMAG]], ptr %[[RESULT_IMAG_PTR]], align 4
Loading