Skip to content

[CIR] Complex Unary plus and minus with promoted type #155486

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

Jump to bottom
Merged
merged 2 commits into from
Aug 29, 2025
Merged

[CIR] Complex Unary plus and minus with promoted type #155486

merged 2 commits into from
Aug 29, 2025

Conversation

AmrDeveloper
Copy link
Member

This change adds support for Complex unary plus and minus expressions with promoted type

Issue: #141365

@llvmbot llvmbot added clang Clang issues not falling into any other category ClangIR Anything related to the ClangIR project labels Aug 26, 2025
@llvmbot
Copy link
Member

llvmbot commented Aug 26, 2025

@llvm/pr-subscribers-clangir

Author: Amr Hesham (AmrDeveloper)

Changes

This change adds support for Complex unary plus and minus expressions with promoted type

Issue: #141365

Full diff: https://github.com/llvm/llvm-project/pull/155486.diff

2 Files Affected:

  • (modified) clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp (+4-8)
  • (modified) clang/test/CIR/CodeGen/complex-unary.cpp (+118)
diff --git a/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp b/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
index bb1b55f2d16f4..a86b2d5e6ed1b 100644
--- a/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
+++ b/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
@@ -323,10 +323,8 @@ mlir::Value ComplexExprEmitter::emitCast(CastKind ck, Expr *op,
 mlir::Value ComplexExprEmitter::VisitUnaryPlus(const UnaryOperator *e) {
   QualType promotionTy = getPromotionType(e->getSubExpr()->getType());
   mlir::Value result = VisitPlusMinus(e, cir::UnaryOpKind::Plus, promotionTy);
-  if (!promotionTy.isNull()) {
-    cgf.cgm.errorNYI("ComplexExprEmitter::VisitUnaryPlus emitUnPromotedValue");
-    return {};
-  }
+  if (!promotionTy.isNull())
+    return cgf.emitUnPromotedValue(result, e->getSubExpr()->getType());
   return result;
 }
 
@@ -348,10 +346,8 @@ mlir::Value ComplexExprEmitter::VisitPlusMinus(const UnaryOperator *e,
 mlir::Value ComplexExprEmitter::VisitUnaryMinus(const UnaryOperator *e) {
   QualType promotionTy = getPromotionType(e->getSubExpr()->getType());
   mlir::Value result = VisitPlusMinus(e, cir::UnaryOpKind::Minus, promotionTy);
-  if (!promotionTy.isNull()) {
-    cgf.cgm.errorNYI("ComplexExprEmitter::VisitUnaryMinus emitUnPromotedValue");
-    return {};
-  }
+  if (!promotionTy.isNull())
+    return cgf.emitUnPromotedValue(result, e->getSubExpr()->getType());
   return result;
 }
 
diff --git a/clang/test/CIR/CodeGen/complex-unary.cpp b/clang/test/CIR/CodeGen/complex-unary.cpp
index 4cd81eb40597a..81803817e0c99 100644
--- a/clang/test/CIR/CodeGen/complex-unary.cpp
+++ b/clang/test/CIR/CodeGen/complex-unary.cpp
@@ -370,3 +370,121 @@ void foo8() {
 // OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[B_ADDR]], i32 0, i32 1
 // OGCG: store float %[[A_REAL_MINUS]], ptr %[[B_REAL_PTR]], align 4
 // OGCG: store float %[[A_IMAG_MINUS]], ptr %[[B_IMAG_PTR]], align 4
+
+void foo9() {
+  _Float16 _Complex a;
+  _Float16 _Complex b = +a;
+}
+
+// CIR: %[[A_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["a"]
+// CIR: %[[B_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["b", init]
+// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.complex<!cir.f16>>, !cir.complex<!cir.f16>
+// CIR: %[[A_REAL:.*]] = cir.complex.real %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_IMAG:.*]] = cir.complex.imag %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_REAL_F32:.*]] = cir.cast(floating, %[[A_REAL]] : !cir.f16), !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.cast(floating, %[[A_IMAG]] : !cir.f16), !cir.float
+// CIR: %[[A_COMPLEX_F32:.*]] = cir.complex.create %[[A_REAL_F32]], %[[A_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[A_REAL_F32:.*]] = cir.complex.real %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.complex.imag %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.unary(plus, %[[A_REAL_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.unary(plus, %[[A_IMAG_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_COMPLEX_F32:.*]] = cir.complex.create %[[RESULT_REAL_F32]], %[[RESULT_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.complex.real %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.complex.imag %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F16:.*]] = cir.cast(floating, %[[RESULT_REAL_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_IMAG_F16:.*]] = cir.cast(floating, %[[RESULT_IMAG_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_COMPLEX_F16:.*]] = cir.complex.create %[[RESULT_REAL_F16]], %[[RESULT_IMAG_F16]] : !cir.f16 -> !cir.complex<!cir.f16>
+// CIR: cir.store{{.*}} %[[RESULT_COMPLEX_F16]], %[[B_ADDR]] : !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>
+
+// LLVM: %[[A_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[B_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[TMP_A:.*]] = load { half, half }, ptr %[[A_ADDR]], align 2
+// LLVM: %[[A_REAL:.*]] = extractvalue { half, half } %[[TMP_A]], 0
+// LLVM: %[[A_IMAG:.*]] = extractvalue { half, half } %[[TMP_A]], 1
+// LLVM: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// LLVM: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[A_REAL_F16:.*]] = fptrunc float %[[A_REAL_F32]] to half
+// LLVM: %[[A_IMAG_F16:.*]] = fptrunc float %[[A_IMAG_F32]] to half
+// LLVM: %[[TMP_RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } {{.*}}, half %[[A_REAL_F16]], 0
+// LLVM: %[[RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } %[[TMP_RESULT_COMPLEX_F16]], half %[[A_IMAG_F16]], 1
+// LLVM: store { half, half } %[[RESULT_COMPLEX_F16]], ptr %[[B_ADDR]], align 2
+
+// OGCG: %[[A_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[B_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 0
+// OGCG: %[[A_REAL:.*]] = load half, ptr %a.realp, align 2
+// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 1
+// OGCG: %[[A_IMAG:.*]] = load half, ptr %a.imagp, align 2
+// OGCG: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// OGCG: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// OGCG: %[[RESULT_REAL:.*]] = fptrunc float %[[A_REAL_F32]] to half
+// OGCG: %[[RESULT_IMAG:.*]] = fptrunc float %[[A_IMAG_F32]] to half
+// OGCG: %[[B_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 0
+// OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 1
+// OGCG: store half %[[RESULT_REAL]], ptr %[[B_REAL_PTR]], align 2
+// OGCG: store half %[[RESULT_IMAG]], ptr %[[B_IMAG_PTR]], align 2
+
+void foo10() {
+  _Float16 _Complex a;
+  _Float16 _Complex b = -a;
+}
+
+// CIR: %[[A_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["a"]
+// CIR: %[[B_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["b", init]
+// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.complex<!cir.f16>>, !cir.complex<!cir.f16>
+// CIR: %[[A_REAL:.*]] = cir.complex.real %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_IMAG:.*]] = cir.complex.imag %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_REAL_F32:.*]] = cir.cast(floating, %[[A_REAL]] : !cir.f16), !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.cast(floating, %[[A_IMAG]] : !cir.f16), !cir.float
+// CIR: %[[A_COMPLEX_F32:.*]] = cir.complex.create %[[A_REAL_F32]], %[[A_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[A_REAL_F32:.*]] = cir.complex.real %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.complex.imag %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.unary(minus, %[[A_REAL_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.unary(minus, %[[A_IMAG_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_COMPLEX_F32:.*]] = cir.complex.create %[[RESULT_REAL_F32]], %[[RESULT_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.complex.real %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.complex.imag %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F16:.*]] = cir.cast(floating, %[[RESULT_REAL_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_IMAG_F16:.*]] = cir.cast(floating, %[[RESULT_IMAG_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_COMPLEX_F16:.*]] = cir.complex.create %[[RESULT_REAL_F16]], %[[RESULT_IMAG_F16]] : !cir.f16 -> !cir.complex<!cir.f16>
+// CIR: cir.store{{.*}} %[[RESULT_COMPLEX_F16]], %[[B_ADDR]] : !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>
+
+// LLVM: %[[A_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[B_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[TMP_A:.*]] = load { half, half }, ptr %[[A_ADDR]], align 2
+// LLVM: %[[A_REAL:.*]] = extractvalue { half, half } %[[TMP_A]], 0
+// LLVM: %[[A_IMAG:.*]] = extractvalue { half, half } %[[TMP_A]], 1
+// LLVM: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// LLVM: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[RESULT_REAL_F32:.*]] = fneg float %[[A_REAL_F32]]
+// LLVM: %[[RESULT_IMAG_F32:.*]] = fneg float %[[A_IMAG_F32]]
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[RESULT_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[RESULT_IMAG_F32]], 1
+// LLVM: %[[A_REAL_F16:.*]] = fptrunc float %[[RESULT_REAL_F32]] to half
+// LLVM: %[[A_IMAG_F16:.*]] = fptrunc float %[[RESULT_IMAG_F32]] to half
+// LLVM: %[[TMP_RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } {{.*}}, half %[[A_REAL_F16]], 0
+// LLVM: %[[RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } %[[TMP_RESULT_COMPLEX_F16]], half %[[A_IMAG_F16]], 1
+// LLVM: store { half, half } %[[RESULT_COMPLEX_F16]], ptr %[[B_ADDR]], align 2
+
+// OGCG: %[[A_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[B_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 0
+// OGCG: %[[A_REAL:.*]] = load half, ptr %a.realp, align 2
+// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 1
+// OGCG: %[[A_IMAG:.*]] = load half, ptr %a.imagp, align 2
+// OGCG: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// OGCG: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// OGCG: %[[RESULT_REAL_F32:.*]] = fneg float %[[A_REAL_F32]]
+// OGCG: %[[RESULT_IMAG_F32:.*]] = fneg float %[[A_IMAG_F32]]
+// OGCG: %[[RESULT_REAL:.*]] = fptrunc float %[[RESULT_REAL_F32]] to half
+// OGCG: %[[RESULT_IMAG:.*]] = fptrunc float %[[RESULT_IMAG_F32]] to half
+// OGCG: %[[B_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 0
+// OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 1
+// OGCG: store half %[[RESULT_REAL]], ptr %[[B_REAL_PTR]], align 2
+// OGCG: store half %[[RESULT_IMAG]], ptr %[[B_IMAG_PTR]], align 2

@llvmbot
Copy link
Member

llvmbot commented Aug 26, 2025

@llvm/pr-subscribers-clang

Author: Amr Hesham (AmrDeveloper)

Changes

This change adds support for Complex unary plus and minus expressions with promoted type

Issue: #141365

Full diff: https://github.com/llvm/llvm-project/pull/155486.diff

2 Files Affected:

  • (modified) clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp (+4-8)
  • (modified) clang/test/CIR/CodeGen/complex-unary.cpp (+118)
diff --git a/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp b/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
index bb1b55f2d16f4..a86b2d5e6ed1b 100644
--- a/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
+++ b/clang/lib/CIR/CodeGen/CIRGenExprComplex.cpp
@@ -323,10 +323,8 @@ mlir::Value ComplexExprEmitter::emitCast(CastKind ck, Expr *op,
 mlir::Value ComplexExprEmitter::VisitUnaryPlus(const UnaryOperator *e) {
   QualType promotionTy = getPromotionType(e->getSubExpr()->getType());
   mlir::Value result = VisitPlusMinus(e, cir::UnaryOpKind::Plus, promotionTy);
-  if (!promotionTy.isNull()) {
-    cgf.cgm.errorNYI("ComplexExprEmitter::VisitUnaryPlus emitUnPromotedValue");
-    return {};
-  }
+  if (!promotionTy.isNull())
+    return cgf.emitUnPromotedValue(result, e->getSubExpr()->getType());
   return result;
 }
 
@@ -348,10 +346,8 @@ mlir::Value ComplexExprEmitter::VisitPlusMinus(const UnaryOperator *e,
 mlir::Value ComplexExprEmitter::VisitUnaryMinus(const UnaryOperator *e) {
   QualType promotionTy = getPromotionType(e->getSubExpr()->getType());
   mlir::Value result = VisitPlusMinus(e, cir::UnaryOpKind::Minus, promotionTy);
-  if (!promotionTy.isNull()) {
-    cgf.cgm.errorNYI("ComplexExprEmitter::VisitUnaryMinus emitUnPromotedValue");
-    return {};
-  }
+  if (!promotionTy.isNull())
+    return cgf.emitUnPromotedValue(result, e->getSubExpr()->getType());
   return result;
 }
 
diff --git a/clang/test/CIR/CodeGen/complex-unary.cpp b/clang/test/CIR/CodeGen/complex-unary.cpp
index 4cd81eb40597a..81803817e0c99 100644
--- a/clang/test/CIR/CodeGen/complex-unary.cpp
+++ b/clang/test/CIR/CodeGen/complex-unary.cpp
@@ -370,3 +370,121 @@ void foo8() {
 // OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { float, float }, ptr %[[B_ADDR]], i32 0, i32 1
 // OGCG: store float %[[A_REAL_MINUS]], ptr %[[B_REAL_PTR]], align 4
 // OGCG: store float %[[A_IMAG_MINUS]], ptr %[[B_IMAG_PTR]], align 4
+
+void foo9() {
+  _Float16 _Complex a;
+  _Float16 _Complex b = +a;
+}
+
+// CIR: %[[A_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["a"]
+// CIR: %[[B_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["b", init]
+// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.complex<!cir.f16>>, !cir.complex<!cir.f16>
+// CIR: %[[A_REAL:.*]] = cir.complex.real %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_IMAG:.*]] = cir.complex.imag %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_REAL_F32:.*]] = cir.cast(floating, %[[A_REAL]] : !cir.f16), !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.cast(floating, %[[A_IMAG]] : !cir.f16), !cir.float
+// CIR: %[[A_COMPLEX_F32:.*]] = cir.complex.create %[[A_REAL_F32]], %[[A_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[A_REAL_F32:.*]] = cir.complex.real %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.complex.imag %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.unary(plus, %[[A_REAL_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.unary(plus, %[[A_IMAG_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_COMPLEX_F32:.*]] = cir.complex.create %[[RESULT_REAL_F32]], %[[RESULT_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.complex.real %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.complex.imag %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F16:.*]] = cir.cast(floating, %[[RESULT_REAL_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_IMAG_F16:.*]] = cir.cast(floating, %[[RESULT_IMAG_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_COMPLEX_F16:.*]] = cir.complex.create %[[RESULT_REAL_F16]], %[[RESULT_IMAG_F16]] : !cir.f16 -> !cir.complex<!cir.f16>
+// CIR: cir.store{{.*}} %[[RESULT_COMPLEX_F16]], %[[B_ADDR]] : !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>
+
+// LLVM: %[[A_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[B_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[TMP_A:.*]] = load { half, half }, ptr %[[A_ADDR]], align 2
+// LLVM: %[[A_REAL:.*]] = extractvalue { half, half } %[[TMP_A]], 0
+// LLVM: %[[A_IMAG:.*]] = extractvalue { half, half } %[[TMP_A]], 1
+// LLVM: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// LLVM: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[A_REAL_F16:.*]] = fptrunc float %[[A_REAL_F32]] to half
+// LLVM: %[[A_IMAG_F16:.*]] = fptrunc float %[[A_IMAG_F32]] to half
+// LLVM: %[[TMP_RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } {{.*}}, half %[[A_REAL_F16]], 0
+// LLVM: %[[RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } %[[TMP_RESULT_COMPLEX_F16]], half %[[A_IMAG_F16]], 1
+// LLVM: store { half, half } %[[RESULT_COMPLEX_F16]], ptr %[[B_ADDR]], align 2
+
+// OGCG: %[[A_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[B_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 0
+// OGCG: %[[A_REAL:.*]] = load half, ptr %a.realp, align 2
+// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 1
+// OGCG: %[[A_IMAG:.*]] = load half, ptr %a.imagp, align 2
+// OGCG: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// OGCG: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// OGCG: %[[RESULT_REAL:.*]] = fptrunc float %[[A_REAL_F32]] to half
+// OGCG: %[[RESULT_IMAG:.*]] = fptrunc float %[[A_IMAG_F32]] to half
+// OGCG: %[[B_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 0
+// OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 1
+// OGCG: store half %[[RESULT_REAL]], ptr %[[B_REAL_PTR]], align 2
+// OGCG: store half %[[RESULT_IMAG]], ptr %[[B_IMAG_PTR]], align 2
+
+void foo10() {
+  _Float16 _Complex a;
+  _Float16 _Complex b = -a;
+}
+
+// CIR: %[[A_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["a"]
+// CIR: %[[B_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["b", init]
+// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.complex<!cir.f16>>, !cir.complex<!cir.f16>
+// CIR: %[[A_REAL:.*]] = cir.complex.real %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_IMAG:.*]] = cir.complex.imag %[[TMP_A]] : !cir.complex<!cir.f16> -> !cir.f16
+// CIR: %[[A_REAL_F32:.*]] = cir.cast(floating, %[[A_REAL]] : !cir.f16), !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.cast(floating, %[[A_IMAG]] : !cir.f16), !cir.float
+// CIR: %[[A_COMPLEX_F32:.*]] = cir.complex.create %[[A_REAL_F32]], %[[A_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[A_REAL_F32:.*]] = cir.complex.real %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[A_IMAG_F32:.*]] = cir.complex.imag %[[A_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.unary(minus, %[[A_REAL_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.unary(minus, %[[A_IMAG_F32:.*]]) : !cir.float, !cir.float
+// CIR: %[[RESULT_COMPLEX_F32:.*]] = cir.complex.create %[[RESULT_REAL_F32]], %[[RESULT_IMAG_F32]] : !cir.float -> !cir.complex<!cir.float>
+// CIR: %[[RESULT_REAL_F32:.*]] = cir.complex.real %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_IMAG_F32:.*]] = cir.complex.imag %[[RESULT_COMPLEX_F32]] : !cir.complex<!cir.float> -> !cir.float
+// CIR: %[[RESULT_REAL_F16:.*]] = cir.cast(floating, %[[RESULT_REAL_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_IMAG_F16:.*]] = cir.cast(floating, %[[RESULT_IMAG_F32]] : !cir.float), !cir.f16
+// CIR: %[[RESULT_COMPLEX_F16:.*]] = cir.complex.create %[[RESULT_REAL_F16]], %[[RESULT_IMAG_F16]] : !cir.f16 -> !cir.complex<!cir.f16>
+// CIR: cir.store{{.*}} %[[RESULT_COMPLEX_F16]], %[[B_ADDR]] : !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>
+
+// LLVM: %[[A_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[B_ADDR:.*]] = alloca { half, half }, i64 1, align 2
+// LLVM: %[[TMP_A:.*]] = load { half, half }, ptr %[[A_ADDR]], align 2
+// LLVM: %[[A_REAL:.*]] = extractvalue { half, half } %[[TMP_A]], 0
+// LLVM: %[[A_IMAG:.*]] = extractvalue { half, half } %[[TMP_A]], 1
+// LLVM: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// LLVM: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[A_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[A_IMAG_F32]], 1
+// LLVM: %[[RESULT_REAL_F32:.*]] = fneg float %[[A_REAL_F32]]
+// LLVM: %[[RESULT_IMAG_F32:.*]] = fneg float %[[A_IMAG_F32]]
+// LLVM: %[[TMP_A_COMPLEX_F32:.*]] = insertvalue { float, float } {{.*}}, float %[[RESULT_REAL_F32]], 0
+// LLVM: %[[A_COMPLEX_F32:.*]] = insertvalue { float, float } %[[TMP_A_COMPLEX_F32]], float %[[RESULT_IMAG_F32]], 1
+// LLVM: %[[A_REAL_F16:.*]] = fptrunc float %[[RESULT_REAL_F32]] to half
+// LLVM: %[[A_IMAG_F16:.*]] = fptrunc float %[[RESULT_IMAG_F32]] to half
+// LLVM: %[[TMP_RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } {{.*}}, half %[[A_REAL_F16]], 0
+// LLVM: %[[RESULT_COMPLEX_F16:.*]] = insertvalue { half, half } %[[TMP_RESULT_COMPLEX_F16]], half %[[A_IMAG_F16]], 1
+// LLVM: store { half, half } %[[RESULT_COMPLEX_F16]], ptr %[[B_ADDR]], align 2
+
+// OGCG: %[[A_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[B_ADDR:.*]] = alloca { half, half }, align 2
+// OGCG: %[[A_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 0
+// OGCG: %[[A_REAL:.*]] = load half, ptr %a.realp, align 2
+// OGCG: %[[A_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[A_ADDR]], i32 0, i32 1
+// OGCG: %[[A_IMAG:.*]] = load half, ptr %a.imagp, align 2
+// OGCG: %[[A_REAL_F32:.*]] = fpext half %[[A_REAL]] to float
+// OGCG: %[[A_IMAG_F32:.*]] = fpext half %[[A_IMAG]] to float
+// OGCG: %[[RESULT_REAL_F32:.*]] = fneg float %[[A_REAL_F32]]
+// OGCG: %[[RESULT_IMAG_F32:.*]] = fneg float %[[A_IMAG_F32]]
+// OGCG: %[[RESULT_REAL:.*]] = fptrunc float %[[RESULT_REAL_F32]] to half
+// OGCG: %[[RESULT_IMAG:.*]] = fptrunc float %[[RESULT_IMAG_F32]] to half
+// OGCG: %[[B_REAL_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 0
+// OGCG: %[[B_IMAG_PTR:.*]] = getelementptr inbounds nuw { half, half }, ptr %[[B_ADDR]], i32 0, i32 1
+// OGCG: store half %[[RESULT_REAL]], ptr %[[B_REAL_PTR]], align 2
+// OGCG: store half %[[RESULT_IMAG]], ptr %[[B_IMAG_PTR]], align 2

andykaylor
andykaylor reviewed Aug 27, 2025
View reviewed changes
clang/test/CIR/CodeGen/complex-unary.cpp Outdated
_Float16 _Complex b = +a;
}

// CIR: %[[A_ADDR:.*]] = cir.alloca !cir.complex<!cir.f16>, !cir.ptr<!cir.complex<!cir.f16>>, ["a"]
Copy link
Contributor

Choose a reason for hiding this comment

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

Can you add CIR-BEFORE checks here? I was hoping to see just a complex unary op and a single cast. Is that what we get before LoweringPrepare?

Copy link
Member Author

Choose a reason for hiding this comment

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

Before LP, there were

Cast (Complex<f16> -> Complex<f32>)
Unary (Complex<f32>)
Cast (Complex<f32> -> Complex<f16>)

I will add a test for LP before now

andykaylor
andykaylor approved these changes Aug 29, 2025
View reviewed changes
Copy link
Contributor

@andykaylor andykaylor left a comment

Choose a reason for hiding this comment

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

lgtm

@AmrDeveloper AmrDeveloper merged commit a871207 into llvm:main Aug 29, 2025
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@bcardosolopes bcardosolopes bcardosolopes approved these changes

@andykaylor andykaylor andykaylor approved these changes

@lanza lanza Awaiting requested review from lanza lanza is a code owner

@xlauko xlauko Awaiting requested review from xlauko xlauko is a code owner

Assignees
No one assigned
Labels
clang Clang issues not falling into any other category ClangIR Anything related to the ClangIR project
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@AmrDeveloper @llvmbot @bcardosolopes @andykaylor