[clang] d6db026 - [CIR] Upstream missing support for floating point unary operator (#193215)

via cfe-commits cfe-commits at lists.llvm.org
Thu Oct 1 21:50:54 PDT 2026


Author: Haocong Lu
Date: 2026-10-02T12:50:47+08:00
New Revision: d6db026f11dfed121bde1820d443542adebdaad3

URL: https://github.com/llvm/llvm-project/commit/d6db026f11dfed121bde1820d443542adebdaad3
DIFF: https://github.com/llvm/llvm-project/commit/d6db026f11dfed121bde1820d443542adebdaad3.diff

LOG: [CIR] Upstream missing support for floating point unary operator (#193215)

- Add Support for `__bf16/Float16/bfloat16/float128` scalar pre/post
   increment and decrement.
- Add support for float vector increment and decrement, and add test
  cases for both float vector and integer vector.
- Add test case for __bf16/Float16/bfloat16/double/long double/
  float128 scalar unary operator with target `x86_64`.
- Part of task https://github.com/llvm/llvm-project/issues/192316

Added: 
    

Modified: 
    clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp
    clang/lib/CIR/Lowering/DirectToLLVM/LowerToLLVM.cpp
    clang/lib/CodeGen/CGExprScalar.cpp
    clang/test/CIR/CodeGen/long-double-inc-dec.cpp
    clang/test/CIR/CodeGen/unary.cpp
    clang/test/CIR/CodeGenOpenACC/atomic-update.cpp

Removed: 
    


################################################################################
diff  --git a/clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp b/clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp
index 80e8b36dfda6b..6cae1200e6e23 100644
--- a/clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp
+++ b/clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp
@@ -685,7 +685,7 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
             lv.isVolatileQualified(), /*fetch_first=*/true);
         mlir::Value oldVal = rmw->getResult(0);
         // Prefix returns new value; postfix returns old value.
-        return isPre ? emitIncOrDec(e, oldVal) : oldVal;
+        return isPre ? emitIntIncOrDec(e, oldVal) : oldVal;
       }
 
       // Special case for atomic increment/decrement on floats.
@@ -767,7 +767,7 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
         value = emitIncDecConsiderOverflowBehavior(e, value);
       } else {
         // NOTE(CIR): clang calls CreateAdd but folds this to a unary op
-        value = emitIncOrDec(e, input, /*nsw=*/false);
+        value = emitIntIncOrDec(e, input, /*nsw=*/false);
       }
     } else if (const PointerType *ptr = type->getAs<PointerType>()) {
       QualType type = ptr->getPointeeType();
@@ -789,32 +789,14 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
       }
     } else if (type->isVectorType()) {
       if (type->hasIntegerRepresentation()) {
-        value = emitIncOrDec(e, input, /*nsw=*/false);
+        value = emitIntIncOrDec(e, input, /*nsw=*/false);
       } else {
-        cgf.cgm.errorNYI(e->getSourceRange(), "Unary inc/dec vector of float");
-        return {};
+        CIRGenFunction::CIRGenFPOptionsRAII FPOptsRAII(cgf, e);
+        value = emitFloatIncOrDec(e, input);
       }
     } else if (type->isRealFloatingType()) {
       CIRGenFunction::CIRGenFPOptionsRAII FPOptsRAII(cgf, e);
-
-      if (type->isHalfType() &&
-          !cgf.getContext().getLangOpts().NativeHalfType) {
-        cgf.cgm.errorNYI(e->getSourceRange(), "Unary inc/dec half");
-        return {};
-      }
-
-      if (mlir::isa<cir::SingleType, cir::DoubleType, cir::LongDoubleType>(
-              value.getType())) {
-        mlir::Location loc = cgf.getLoc(e->getExprLoc());
-        auto fpType = mlir::cast<cir::FPTypeInterface>(value.getType());
-        mlir::Value amount = builder.getConstFP(
-            loc, value.getType(), llvm::APFloat(fpType.getFloatSemantics(), 1));
-        value = e->isIncrementOp() ? builder.createFAdd(loc, value, amount)
-                                   : builder.createFSub(loc, value, amount);
-      } else {
-        cgf.cgm.errorNYI(e->getSourceRange(), "Unary inc/dec other fp type");
-        return {};
-      }
+      value = emitFloatIncOrDec(e, input);
     } else if (type->isFixedPointType()) {
       value = emitFixedPointIncDec(e, value, type);
     } else {
@@ -840,13 +822,13 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
                                                  mlir::Value inVal) {
     switch (cgf.getLangOpts().getSignedOverflowBehavior()) {
     case LangOptions::SOB_Defined:
-      return emitIncOrDec(e, inVal, /*nsw=*/false);
+      return emitIntIncOrDec(e, inVal, /*nsw=*/false);
     case LangOptions::SOB_Undefined:
       assert(!cir::MissingFeatures::sanitizers());
-      return emitIncOrDec(e, inVal, /*nsw=*/true);
+      return emitIntIncOrDec(e, inVal, /*nsw=*/true);
     case LangOptions::SOB_Trapping:
       if (!e->canOverflow())
-        return emitIncOrDec(e, inVal, /*nsw=*/true);
+        return emitIntIncOrDec(e, inVal, /*nsw=*/true);
       cgf.cgm.errorNYI(e->getSourceRange(), "inc/def overflow SOB_Trapping");
       return {};
     }
@@ -912,14 +894,46 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
     return builder.createOrFold<cir::MinusOp>(loc, operand, nsw);
   }
 
-  mlir::Value emitIncOrDec(const UnaryOperator *e, mlir::Value input,
-                           bool nsw = false) {
+  mlir::Value emitIntIncOrDec(const UnaryOperator *e, mlir::Value input,
+                              bool nsw = false) {
     mlir::Location loc = cgf.getLoc(e->getSourceRange().getBegin());
     return e->isIncrementOp()
                ? builder.createOrFold<cir::IncOp>(loc, input, nsw)
                : builder.createOrFold<cir::DecOp>(loc, input, nsw);
   }
 
+  mlir::Value emitFloatIncOrDec(const UnaryOperator *e, mlir::Value input) {
+    assert(cir::isFPOrVectorOfFPType(input.getType()) &&
+           "Expect floating-point operand");
+    mlir::Location loc = cgf.getLoc(e->getSourceRange().getBegin());
+
+    if (auto vecType = mlir::dyn_cast<cir::VectorType>(input.getType())) {
+      mlir::Type fpScalarType = vecType.getElementType();
+      auto fpInterface = mlir::cast<cir::FPTypeInterface>(fpScalarType);
+      auto amount = llvm::APFloat::getOne(fpInterface.getFloatSemantics(),
+                                          /*Negative=*/e->isDecrementOp());
+      mlir::Value amtValue = builder.getConstFP(loc, fpScalarType, amount);
+      amtValue = cir::VecSplatOp::create(builder, loc, vecType, amtValue);
+      return builder.createFAdd(loc, input, amtValue);
+    }
+
+    QualType type = e->getSubExpr()->getType();
+    // Another special case: half FP increment should be done via float.
+    if (type->isHalfType() && !cgf.getContext().getLangOpts().NativeHalfType)
+      input = builder.createFloatingCast(input, builder.getSingleTy());
+
+    auto fpInterface = mlir::cast<cir::FPTypeInterface>(input.getType());
+    auto amount = llvm::APFloat::getOne(fpInterface.getFloatSemantics(),
+                                        /*Negative=*/e->isDecrementOp());
+    mlir::Value amtValue = builder.getConstFP(loc, input.getType(), amount);
+    mlir::Value output = builder.createFAdd(loc, input, amtValue);
+
+    if (type->isHalfType() && !cgf.getContext().getLangOpts().NativeHalfType)
+      output = builder.createFloatingCast(output, builder.getFp16Ty());
+
+    return output;
+  }
+
   mlir::Value VisitUnaryNot(const UnaryOperator *e) {
     ignoreResultAssign = false;
     mlir::Value op = Visit(e->getSubExpr());

diff  --git a/clang/lib/CIR/Lowering/DirectToLLVM/LowerToLLVM.cpp b/clang/lib/CIR/Lowering/DirectToLLVM/LowerToLLVM.cpp
index e678a33e98a94..007b673419056 100644
--- a/clang/lib/CIR/Lowering/DirectToLLVM/LowerToLLVM.cpp
+++ b/clang/lib/CIR/Lowering/DirectToLLVM/LowerToLLVM.cpp
@@ -3313,8 +3313,9 @@ lowerIncDecOp(CIROp op, typename CIROp::Adaptor adaptor,
   auto maybeNSW = nswFlag(op.getNoSignedWrap());
   mlir::LLVM::ConstantOp one;
   if (mlir::isa<cir::VectorType>(op.getType())) {
-    mlir::DenseIntElementsAttr oneVec = mlir::DenseIntElementsAttr::get(
-        mlir::cast<mlir::ShapedType>(llvmType), 1);
+    mlir::ShapedType shapedTy = mlir::cast<mlir::ShapedType>(llvmType);
+    mlir::APInt oneAP(shapedTy.getElementTypeBitWidth(), 1);
+    auto oneVec = mlir::DenseElementsAttr::get(shapedTy, {oneAP});
     one = mlir::LLVM::ConstantOp::create(rewriter, loc, llvmType, oneVec);
   } else {
     one = mlir::LLVM::ConstantOp::create(rewriter, loc, llvmType, 1);

diff  --git a/clang/lib/CodeGen/CGExprScalar.cpp b/clang/lib/CodeGen/CGExprScalar.cpp
index a024562b82f20..24c6b6c0c1ece 100644
--- a/clang/lib/CodeGen/CGExprScalar.cpp
+++ b/clang/lib/CodeGen/CGExprScalar.cpp
@@ -3554,12 +3554,9 @@ ScalarExprEmitter::EmitScalarPrePostIncDec(const UnaryOperator *E, LValue LV,
     llvm::Value *amt;
     CodeGenFunction::CGFPOptionsRAII FPOptsRAII(CGF, E);
 
-    if (type->isHalfType() && !CGF.getContext().getLangOpts().NativeHalfType) {
-      // Another special case: half FP increment should be done via float. If
-      // the input isn't already half, it may be i16.
-      Value *bitcast = Builder.CreateBitCast(input, CGF.CGM.HalfTy);
-      value = Builder.CreateFPExt(bitcast, CGF.CGM.FloatTy, "incdec.conv");
-    }
+    // Another special case: half FP increment should be done via float.
+    if (type->isHalfType() && !CGF.getContext().getLangOpts().NativeHalfType)
+      value = Builder.CreateFPExt(value, CGF.CGM.FloatTy, "incdec.conv");
 
     if (value->getType()->isFloatTy())
       amt = llvm::ConstantFP::get(VMContext,
@@ -3590,12 +3587,10 @@ ScalarExprEmitter::EmitScalarPrePostIncDec(const UnaryOperator *E, LValue LV,
     }
     value = Builder.CreateFAdd(value, amt, isInc ? "inc" : "dec");
 
-    if (type->isHalfType() && !CGF.getContext().getLangOpts().NativeHalfType) {
+    if (type->isHalfType() && !CGF.getContext().getLangOpts().NativeHalfType)
       value = Builder.CreateFPTrunc(value, CGF.CGM.HalfTy, "incdec.conv");
-      value = Builder.CreateBitCast(value, input->getType());
-    }
 
-  // Fixed-point types.
+    // Fixed-point types.
   } else if (type->isFixedPointType()) {
     // Fixed-point types are tricky. In some cases, it isn't possible to
     // represent a 1 or a -1 in the type at all. Piggyback off of

diff  --git a/clang/test/CIR/CodeGen/long-double-inc-dec.cpp b/clang/test/CIR/CodeGen/long-double-inc-dec.cpp
index daf71e5af5a46..4321a63c0fbc9 100644
--- a/clang/test/CIR/CodeGen/long-double-inc-dec.cpp
+++ b/clang/test/CIR/CodeGen/long-double-inc-dec.cpp
@@ -38,8 +38,7 @@ extern "C" long double do_pre_inc(long double d) {
   // CIR: cir.return %[[LOAD_RET]] : !cir.long_double<![[LDTY]]>
   //
   // LLVM: %[[ARG_LOAD:.*]] = load [[LDTY]], ptr %[[ARG_ALLOCA]]
-  // LLVMCIR: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
-  // OGCG: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
+  // LLVM: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
   // LLVM: store [[LDTY]] %[[ARG_INC]], ptr %[[ARG_ALLOCA]]
   // LLVMCIR: store [[LDTY]] %[[ARG_INC]], ptr %[[RET_ALLOCA]]
   // LLVMCIR: %[[LOAD_RET:.*]] = load [[LDTY]], ptr %[[RET_ALLOCA]]
@@ -65,8 +64,7 @@ extern "C" long double do_post_inc(long double d) {
   // CIR: cir.return %[[LOAD_RET]] : !cir.long_double<![[LDTY]]>
   //
   // LLVM: %[[ARG_LOAD:.*]] = load [[LDTY]], ptr %[[ARG_ALLOCA]]
-  // LLVMCIR: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
-  // OGCG: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
+  // LLVM: %[[ARG_INC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
   // LLVM: store [[LDTY]] %[[ARG_INC]], ptr %[[ARG_ALLOCA]]
   // LLVMCIR: store [[LDTY]] %[[ARG_LOAD]], ptr %[[RET_ALLOCA]]
   // LLVMCIR: %[[LOAD_RET:.*]] = load [[LDTY]], ptr %[[RET_ALLOCA]]
@@ -85,15 +83,15 @@ extern "C" long double do_pre_dec(long double d) {
 
   return --d;
   // CIR: %[[ARG_LOAD:.*]]  = cir.load {{.*}}%[[ARG_ALLOCA]] : !cir.ptr<!cir.long_double<![[LDTY]]>>, !cir.long_double<![[LDTY]]>
-  // CIR: %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.long_double<![[LDTY]]>
-  // CIR: %[[ARG_DEC:.*]] = cir.fsub %[[ARG_LOAD]], %[[ONE]]
+  // CIR: %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.long_double<![[LDTY]]>
+  // CIR: %[[ARG_DEC:.*]] = cir.fadd %[[ARG_LOAD]], %[[NEGONE]]
   // CIR: cir.store{{.*}} %[[ARG_DEC]], %[[ARG_ALLOCA]] : !cir.long_double<![[LDTY]]>, !cir.ptr<!cir.long_double<![[LDTY]]>>
   // CIR: cir.store %[[ARG_DEC]], %[[RET_ALLOCA]] : !cir.long_double<![[LDTY]]>, !cir.ptr<!cir.long_double<![[LDTY]]>>
   // CIR: %[[LOAD_RET:.*]] = cir.load %[[RET_ALLOCA]]
   // CIR: cir.return %[[LOAD_RET]] : !cir.long_double<![[LDTY]]>
   //
   // LLVM: %[[ARG_LOAD:.*]] = load [[LDTY]], ptr %[[ARG_ALLOCA]]
-  // LLVMCIR: %[[ARG_DEC:.*]] = fsub [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
+  // LLVMCIR: %[[ARG_DEC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], -1.000000e+00
   // OGCG: %[[ARG_DEC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], -1.000000e+00
   // LLVM: store [[LDTY]] %[[ARG_DEC]], ptr %[[ARG_ALLOCA]]
   // LLVMCIR: store [[LDTY]] %[[ARG_DEC]], ptr %[[RET_ALLOCA]]
@@ -112,15 +110,15 @@ extern "C" long double do_post_dec(long double d) {
 
   return d--;
   // CIR: %[[ARG_LOAD:.*]]  = cir.load {{.*}}%[[ARG_ALLOCA]] : !cir.ptr<!cir.long_double<![[LDTY]]>>, !cir.long_double<![[LDTY]]>
-  // CIR: %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.long_double<![[LDTY]]>
-  // CIR: %[[ARG_DEC:.*]] = cir.fsub %[[ARG_LOAD]], %[[ONE]]
+  // CIR: %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.long_double<![[LDTY]]>
+  // CIR: %[[ARG_DEC:.*]] = cir.fadd %[[ARG_LOAD]], %[[NEGONE]]
   // CIR: cir.store{{.*}} %[[ARG_DEC]], %[[ARG_ALLOCA]] : !cir.long_double<![[LDTY]]>, !cir.ptr<!cir.long_double<![[LDTY]]>>
   // CIR: cir.store %[[ARG_LOAD]], %[[RET_ALLOCA]] : !cir.long_double<![[LDTY]]>, !cir.ptr<!cir.long_double<![[LDTY]]>>
   // CIR: %[[LOAD_RET:.*]] = cir.load %[[RET_ALLOCA]]
   // CIR: cir.return %[[LOAD_RET]] : !cir.long_double<![[LDTY]]>
   //
   // LLVM: %[[ARG_LOAD:.*]] = load [[LDTY]], ptr %[[ARG_ALLOCA]]
-  // LLVMCIR: %[[ARG_DEC:.*]] = fsub [[LDTY]] %[[ARG_LOAD]], 1.000000e+00
+  // LLVMCIR: %[[ARG_DEC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], -1.000000e+00
   // OGCG: %[[ARG_DEC:.*]] = fadd [[LDTY]] %[[ARG_LOAD]], -1.000000e+00
   // LLVM: store [[LDTY]] %[[ARG_DEC]], ptr %[[ARG_ALLOCA]]
   // LLVMCIR: store [[LDTY]] %[[ARG_LOAD]], ptr %[[RET_ALLOCA]]

diff  --git a/clang/test/CIR/CodeGen/unary.cpp b/clang/test/CIR/CodeGen/unary.cpp
index 03b068a68d066..344b0374bd2fb 100644
--- a/clang/test/CIR/CodeGen/unary.cpp
+++ b/clang/test/CIR/CodeGen/unary.cpp
@@ -1,8 +1,8 @@
-// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir -Wno-unused-value -emit-cir %s -o %t.cir
-// RUN: FileCheck --input-file=%t.cir %s
-// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir -Wno-unused-value -emit-llvm %s -o %t-cir.ll
+// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fzvector -fclangir -Wno-unused-value -emit-cir %s -o %t.cir
+// RUN: FileCheck --input-file=%t.cir %s --check-prefix=CIR
+// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fzvector -fclangir -Wno-unused-value -emit-llvm %s -o %t-cir.ll
 // RUN: FileCheck --input-file=%t-cir.ll %s --check-prefix=LLVM
-// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -Wno-unused-value -emit-llvm %s -o %t.ll
+// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fzvector -Wno-unused-value -emit-llvm %s -o %t.ll
 // RUN: FileCheck --input-file=%t.ll %s --check-prefix=OGCG
 
 unsigned up0() {
@@ -10,9 +10,9 @@ unsigned up0() {
   return +a;
 }
 
-// CHECK: cir.func{{.*}} @_Z3up0v() -> (!u32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z3up0v() -> (!u32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} i32 @_Z3up0v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -30,10 +30,10 @@ unsigned um0() {
   return -a;
 }
 
-// CHECK: cir.func{{.*}} @_Z3um0v() -> (!u32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[OUTPUT:.*]] = cir.minus %[[INPUT]]
+// CIR: cir.func{{.*}} @_Z3um0v() -> (!u32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[OUTPUT:.*]] = cir.minus %[[INPUT]]
 
 // LLVM: define{{.*}} i32 @_Z3um0v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -53,10 +53,10 @@ unsigned un0() {
   return ~a; // a ^ -1 , not
 }
 
-// CHECK: cir.func{{.*}} @_Z3un0v() -> (!u32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[OUTPUT:.*]] = cir.not %[[INPUT]]
+// CIR: cir.func{{.*}} @_Z3un0v() -> (!u32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!u32i>
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[OUTPUT:.*]] = cir.not %[[INPUT]]
 
 // LLVM: define{{.*}} i32 @_Z3un0v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -77,14 +77,14 @@ int inc0() {
   return a;
 }
 
-// CHECK: cir.func{{.*}} @_Z4inc0v() -> (!s32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[INCREMENTED:.*]] = cir.inc nsw %[[INPUT]]
-// CHECK:   cir.store{{.*}} %[[INCREMENTED]], %[[A]]
-// CHECK:   %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z4inc0v() -> (!s32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[INCREMENTED:.*]] = cir.inc nsw %[[INPUT]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} i32 @_Z4inc0v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -105,14 +105,14 @@ int dec0() {
   return a;
 }
 
-// CHECK: cir.func{{.*}} @_Z4dec0v() -> (!s32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[DECREMENTED:.*]] = cir.dec nsw %[[INPUT]]
-// CHECK:   cir.store{{.*}} %[[DECREMENTED]], %[[A]]
-// CHECK:   %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z4dec0v() -> (!s32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[DECREMENTED:.*]] = cir.dec nsw %[[INPUT]]
+// CIR:    cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:    %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} i32 @_Z4dec0v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -133,14 +133,14 @@ int inc1() {
   return a;
 }
 
-// CHECK: cir.func{{.*}} @_Z4inc1v() -> (!s32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[INCREMENTED:.*]] = cir.inc nsw %[[INPUT]]
-// CHECK:   cir.store{{.*}} %[[INCREMENTED]], %[[A]]
-// CHECK:   %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z4inc1v() -> (!s32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[INCREMENTED:.*]] = cir.inc nsw %[[INPUT]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} i32 @_Z4inc1v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -161,14 +161,14 @@ int dec1() {
   return a;
 }
 
-// CHECK: cir.func{{.*}} @_Z4dec1v() -> (!s32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[DECREMENTED:.*]] = cir.dec nsw %[[INPUT]]
-// CHECK:   cir.store{{.*}} %[[DECREMENTED]], %[[A]]
-// CHECK:   %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z4dec1v() -> (!s32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[DECREMENTED:.*]] = cir.dec nsw %[[INPUT]]
+// CIR:    cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:    %[[A_TO_OUTPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} i32 @_Z4dec1v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -190,16 +190,16 @@ int inc2() {
   return b;
 }
 
-// CHECK: cir.func{{.*}} @_Z4inc2v() -> (!s32i{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[B:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!s32i>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
-// CHECK:   %[[ATOB:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[INCREMENTED:.*]] = cir.inc nsw %[[ATOB]]
-// CHECK:   cir.store{{.*}} %[[INCREMENTED]], %[[A]]
-// CHECK:   cir.store{{.*}} %[[ATOB]], %[[B]]
-// CHECK:   %[[B_TO_OUTPUT:.*]] = cir.load{{.*}} %[[B]]
+// CIR: cir.func{{.*}} @_Z4inc2v() -> (!s32i{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[B:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!s32i>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.int<1> : !s32i
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !s32i
+// CIR:    %[[ATOB:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[INCREMENTED:.*]] = cir.inc nsw %[[ATOB]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    cir.store{{.*}} %[[ATOB]], %[[B]]
+// CIR:    %[[B_TO_OUTPUT:.*]] = cir.load{{.*}} %[[B]]
 
 // LLVM: define{{.*}} i32 @_Z4inc2v()
 // LLVM:   %[[RV:.*]] = alloca i32, align 4
@@ -222,14 +222,52 @@ int inc2() {
 // OGCG:   store i32 %[[A_LOAD]], ptr %[[B]], align 4
 // OGCG:   %[[B_TO_OUTPUT:.*]] = load i32, ptr %[[B]], align 4
 
+// Chars can go through some integer promotion codegen paths even when not promoted.
+// These should not have nsw attributes because the intermediate promotion makes the
+// overflow defined behavior.
+void chars(char c) {
+  int c1 = +c;
+  int c2 = -c;
+  ++c;
+  --c;
+  c++;
+  c--;
+}
+// CIR: cir.func{{.*}} @_Z5charsc
+// CIR:    %[[PROMO1:.*]] = cir.cast integral %{{.+}} : !s8i -> !s32i
+// CIR:    %[[PROMO2:.*]] = cir.cast integral %{{.+}} : !s8i -> !s32i
+// CIR:    cir.minus nsw %[[PROMO2]] : !s32i
+// CIR:    cir.inc %{{.+}} : !s8i
+// CIR:    cir.dec %{{.+}} : !s8i
+// CIR:    cir.inc %{{.+}} : !s8i
+// CIR:    cir.dec %{{.+}} : !s8i
+
+// LLVM: define{{.*}} void @_Z5charsc
+// LLVM:   %[[PROMO1:.*]] = sext i8 %{{.+}} to i32
+// LLVM:   %[[PROMO2:.*]] = sext i8 %{{.+}} to i32
+// LLVM:   %[[MINUS:.*]] = sub nsw i32 0, %[[PROMO2]]
+// LLVM:   add i8 %{{.+}}, 1
+// LLVM:   sub i8 %{{.+}}, 1
+// LLVM:   add i8 %{{.+}}, 1
+// LLVM:   sub i8 %{{.+}}, 1
+
+// OGCG: define{{.*}} void @_Z5charsc
+// OGCG:   %[[PROMO1:.*]] = sext i8 %{{.+}} to i32
+// OGCG:   %[[PROMO2:.*]] = sext i8 %{{.+}} to i32
+// OGCG:   %[[MINUS:.*]] = sub nsw i32 0, %[[PROMO2]]
+// OGCG:   add i8 %{{.+}}, 1
+// OGCG:   add i8 %{{.+}}, -1
+// OGCG:   add i8 %{{.+}}, 1
+// OGCG:   add i8 %{{.+}}, -1
+
 float fpPlus() {
   float a = 1.0f;
   return +a;
 }
 
-// CHECK: cir.func{{.*}} @_Z6fpPlusv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR: cir.func{{.*}} @_Z6fpPlusv() -> (!cir.float{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
 
 // LLVM: define{{.*}} float @_Z6fpPlusv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
@@ -247,10 +285,10 @@ float fpMinus() {
   return -a;
 }
 
-// CHECK: cir.func{{.*}} @_Z7fpMinusv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[OUTPUT:.*]] = cir.fneg %[[INPUT]]
+// CIR: cir.func{{.*}} @_Z7fpMinusv() -> (!cir.float{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[OUTPUT:.*]] = cir.fneg %[[INPUT]]
 
 // LLVM: define{{.*}} float @_Z7fpMinusv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
@@ -270,104 +308,144 @@ float fpPreInc() {
   return ++a;
 }
 
-// CHECK: cir.func{{.*}} @_Z8fpPreIncv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   %[[INCREMENTED:.*]] = cir.fadd %[[INPUT]], %[[ONE]]
+// CIR: cir.func{{.*}} @_Z8fpPreIncv() -> (!cir.float{{.*}})
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.float>
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    %[[INCREMENTED:.*]] = cir.fadd %[[INPUT]], %[[ONE]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    cir.store %[[INCREMENTED]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.float>, !cir.float
+// CIR:    cir.return %[[RV_LOAD]] : !cir.float
 
 // LLVM: define{{.*}} float @_Z8fpPreIncv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
 // LLVM:   %[[A:.*]] = alloca float, align 4
 // LLVM:   store float 1.000000e+00, ptr %[[A]], align 4
 // LLVM:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// LLVM:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// LLVM:   %[[INCREMENTED:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// LLVM:   store float %[[INCREMENTED]], ptr %[[A]], align 4
+// LLVM:   store float %[[INCREMENTED]], ptr %[[RV]], align 4
+// LLVM:   %[[RV_LOAD:.*]] = load float, ptr %[[RV]], align 4
+// LLVM:   ret float %[[RV_LOAD]]
 
 // OGCG: define{{.*}} float @_Z8fpPreIncv()
 // OGCG:   %[[A:.*]] = alloca float, align 4
 // OGCG:   store float 1.000000e+00, ptr %[[A]], align 4
 // OGCG:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// OGCG:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// OGCG:   %[[INCREMENTED:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// OGCG:   ret float %[[INCREMENTED]]
 
 float fpPreDec() {
   float a = 1.0f;
   return --a;
 }
 
-// CHECK: cir.func{{.*}} @_Z8fpPreDecv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   %[[DECREMENTED:.*]] = cir.fsub %[[INPUT]], %[[ONE]]
+// CIR: cir.func{{.*}} @_Z8fpPreDecv() -> (!cir.float{{.*}})
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.float>
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:    %[[DECREMENTED:.*]] = cir.fadd %[[INPUT]], %[[NEGONE]]
+// CIR:    cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:    cir.store %[[DECREMENTED]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.float>, !cir.float
+// CIR:    cir.return %[[RV_LOAD]] : !cir.float
 
 // LLVM: define{{.*}} float @_Z8fpPreDecv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
 // LLVM:   %[[A:.*]] = alloca float, align 4
 // LLVM:   store float 1.000000e+00, ptr %[[A]], align 4
 // LLVM:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// LLVM:   %[[RESULT:.*]] = fsub float %[[A_LOAD]], 1.000000e+00
+// LLVM:   %[[DECREMENTED:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// LLVM:   store float %[[DECREMENTED]], ptr %[[A]], align 4
+// LLVM:   store float %[[DECREMENTED]], ptr %[[RV]], align 4
+// LLVM:   %[[RV_LOAD:.*]] = load float, ptr %[[RV]], align 4
+// LLVM:   ret float %[[RV_LOAD]]
 
 // OGCG: define{{.*}} float @_Z8fpPreDecv()
 // OGCG:   %[[A:.*]] = alloca float, align 4
 // OGCG:   store float 1.000000e+00, ptr %[[A]], align 4
 // OGCG:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// OGCG:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// OGCG:   %[[DECREMENTED:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// OGCG:   ret float %[[DECREMENTED]]
 
 float fpPostInc() {
   float a = 1.0f;
   return a++;
 }
 
-// CHECK: cir.func{{.*}} @_Z9fpPostIncv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   %[[INCREMENTED:.*]] = cir.fadd %[[INPUT]], %[[ONE]]
+// CIR: cir.func{{.*}} @_Z9fpPostIncv() -> (!cir.float{{.*}})
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.float>
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    %[[INCREMENTED:.*]] = cir.fadd %[[INPUT]], %[[ONE]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    cir.store %[[INPUT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.float>, !cir.float
+// CIR:    cir.return %[[RV_LOAD]] : !cir.float
 
 // LLVM: define{{.*}} float @_Z9fpPostIncv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
 // LLVM:   %[[A:.*]] = alloca float, align 4
 // LLVM:   store float 1.000000e+00, ptr %[[A]], align 4
 // LLVM:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// LLVM:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// LLVM:   %[[INCREMENTED:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// LLVM:   store float %[[INCREMENTED]], ptr %[[A]], align 4
+// LLVM:   store float %[[A_LOAD]], ptr %[[RV]], align 4
+// LLVM:   %[[RV_LOAD:.*]] = load float, ptr %[[RV]], align 4
+// LLVM:   ret float %[[RV_LOAD]]
 
 // OGCG: define{{.*}} float @_Z9fpPostIncv()
 // OGCG:   %[[A:.*]] = alloca float, align 4
 // OGCG:   store float 1.000000e+00, ptr %[[A]], align 4
 // OGCG:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// OGCG:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// OGCG:   %[[INCREMENTED:.*]] = fadd float %[[A_LOAD]], 1.000000e+00
+// OGCG:   ret float %[[A_LOAD]]
 
 float fpPostDec() {
   float a = 1.0f;
   return a--;
 }
 
-// CHECK: cir.func{{.*}} @_Z9fpPostDecv() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   %[[DECREMENTED:.*]] = cir.fsub %[[INPUT]], %[[ONE]]
+// CIR: cir.func{{.*}} @_Z9fpPostDecv() -> (!cir.float{{.*}})
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.float>
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:    %[[DECREMENTED:.*]] = cir.fadd %[[INPUT]], %[[NEGONE]]
+// CIR:    cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:    cir.store %[[INPUT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.float>, !cir.float
+// CIR:    cir.return %[[RV_LOAD]] : !cir.float
 
 // LLVM: define{{.*}} float @_Z9fpPostDecv()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
 // LLVM:   %[[A:.*]] = alloca float, align 4
 // LLVM:   store float 1.000000e+00, ptr %[[A]], align 4
 // LLVM:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// LLVM:   %[[RESULT:.*]] = fsub float %[[A_LOAD]], 1.000000e+00
+// LLVM:   %[[DECREMENTED:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// LLVM:   store float %[[DECREMENTED]], ptr %[[A]], align 4
+// LLVM:   store float %[[A_LOAD]], ptr %[[RV]], align 4
+// LLVM:   %[[RV_LOAD:.*]] = load float, ptr %[[RV]], align 4
+// LLVM:   ret float %[[RV_LOAD]]
 
 // OGCG: define{{.*}} float @_Z9fpPostDecv()
 // OGCG:   %[[A:.*]] = alloca float, align 4
 // OGCG:   store float 1.000000e+00, ptr %[[A]], align 4
 // OGCG:   %[[A_LOAD:.*]] = load float, ptr %[[A]], align 4
-// OGCG:   %[[RESULT:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// OGCG:   %[[DECREMENTED:.*]] = fadd float %[[A_LOAD]], -1.000000e+00
+// OGCG:   ret float %[[A_LOAD]]
 
 // Ensure the increment is performed after the assignment to b.
 float fpPostInc2() {
@@ -376,17 +454,17 @@ float fpPostInc2() {
   return b;
 }
 
-// CHECK: cir.func{{.*}} @_Z10fpPostInc2v() -> (!cir.float{{.*}})
-// CHECK:   %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[B:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.float>
-// CHECK:   %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
-// CHECK:   %[[ATOB:.*]] = cir.load{{.*}} %[[A]]
-// CHECK:   %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-// CHECK:   %[[INCREMENTED:.*]] = cir.fadd %[[ATOB]], %[[ONE]]
-// CHECK:   cir.store{{.*}} %[[INCREMENTED]], %[[A]]
-// CHECK:   cir.store{{.*}} %[[ATOB]], %[[B]]
-// CHECK:   %[[B_TO_OUTPUT:.*]] = cir.load{{.*}} %[[B]]
+// CIR: cir.func{{.*}} @_Z10fpPostInc2v() -> (!cir.float{{.*}})
+// CIR:    %[[A:.*]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[B:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.float>
+// CIR:    %[[ATMP:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    cir.store{{.*}} %[[ATMP]], %[[A]] : !cir.float
+// CIR:    %[[ATOB:.*]] = cir.load{{.*}} %[[A]]
+// CIR:    %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    %[[INCREMENTED:.*]] = cir.fadd %[[ATOB]], %[[ONE]]
+// CIR:    cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:    cir.store{{.*}} %[[ATOB]], %[[B]]
+// CIR:    %[[B_TO_OUTPUT:.*]] = cir.load{{.*}} %[[B]]
 
 // LLVM: define{{.*}} float @_Z10fpPostInc2v()
 // LLVM:   %[[RV:.*]] = alloca float, align 4
@@ -409,65 +487,1094 @@ float fpPostInc2() {
 // OGCG:   store float %[[A_LOAD]], ptr %[[B]], align 4
 // OGCG:   %[[B_TO_OUTPUT:.*]] = load float, ptr %[[B]], align 4
 
-void chars(char c) {
-// CHECK: cir.func{{.*}} @_Z5charsc
+// double unary operations
+double doubleUPlus(double f) {
+  return +f;
+}
 
-  int c1 = +c;
-  // CHECK: %[[PROMO:.*]] = cir.cast integral %{{.+}} : !s8i -> !s32i
-  int c2 = -c;
-  // CHECK: %[[PROMO:.*]] = cir.cast integral %{{.+}} : !s8i -> !s32i
-  // CHECK: cir.minus nsw %[[PROMO]] : !s32i
+// CIR: cir.func{{.*}} @_Z11doubleUPlusd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+
+// LLVM: define{{.*}} double @_Z11doubleUPlusd({{.*}})
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %{{.*}}, align 8
+
+// OGCG: define{{.*}} double @_Z11doubleUPlusd({{.*}})
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %{{.*}}, align 8
+
+double doubleUMinus(double f) {
+  return -f;
+}
+
+// CIR: cir.func{{.*}} @_Z12doubleUMinusd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+// CIR:    %[[DBL_NEGATED:.*]] = cir.fneg %[[DBL_LOAD]]
+
+// LLVM: define{{.*}} double @_Z12doubleUMinusd({{.*}})
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %{{.*}}, align 8
+// LLVM:   %[[DBL_NEGATED:.*]] = fneg double %[[DBL_LOAD]]
 
-  // Chars can go through some integer promotion codegen paths even when not promoted.
-  // These should not have nsw attributes because the intermediate promotion makes the
-  // overflow defined behavior.
-  ++c; // CHECK: cir.inc %{{.+}} : !s8i
-  --c; // CHECK: cir.dec %{{.+}} : !s8i
-  c++; // CHECK: cir.inc %{{.+}} : !s8i
-  c--; // CHECK: cir.dec %{{.+}} : !s8i
+// OGCG: define{{.*}} double @_Z12doubleUMinusd({{.*}})
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %{{.*}}, align 8
+// OGCG:   %[[DBL_NEGATED:.*]] = fneg double %[[DBL_LOAD]]
+
+double doubleUPreInc(double f) {
+  return ++f;
+}
+
+// CIR: cir.func{{.*}} @_Z13doubleUPreIncd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+// CIR:    %[[DBL_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.double
+// CIR:    %[[DBL_INC:.*]] = cir.fadd %[[DBL_LOAD]], %[[DBL_ONE]]
+// CIR:    cir.store{{.*}} %[[DBL_INC]], %[[DBL_F]]
+// CIR:    cir.store %[[DBL_INC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.double>, !cir.double
+// CIR:    cir.return %[[RV_LOAD]] : !cir.double
+
+// LLVM: define{{.*}} double @_Z13doubleUPreIncd({{.*}})
+// LLVM:   %[[DBL_F:.*]] = alloca double, align 8
+// LLVM:   %[[RV:.*]] = alloca double, align 8
+// LLVM:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_INC:.*]] = fadd double %[[DBL_LOAD]], 1.000000e+00
+// LLVM:   store double %[[DBL_INC]], ptr %[[DBL_F]], align 8
+// LLVM:   store double %[[DBL_INC]], ptr %[[RV]], align 8
+// LLVM:   %[[RV_LOAD:.*]] = load double, ptr %[[RV]], align 8
+// LLVM:   ret double %[[RV_LOAD]]
+
+// OGCG: define{{.*}} double @_Z13doubleUPreIncd({{.*}})
+// OGCG:   %[[DBL_F:.*]] = alloca double, align 8
+// OGCG:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_INC:.*]] = fadd double %[[DBL_LOAD]], 1.000000e+00
+// OGCG:   store double %[[DBL_INC]], ptr %[[DBL_F]], align 8
+// OGCG:   ret double %[[DBL_INC]]
+
+double doubleUPreDec(double f) {
+  return --f;
+}
+
+// CIR: cir.func{{.*}} @_Z13doubleUPreDecd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+// CIR:    %[[DBL_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.double
+// CIR:    %[[DBL_DEC:.*]] = cir.fadd %[[DBL_LOAD]], %[[DBL_NEGONE]]
+// CIR:    cir.store{{.*}} %[[DBL_DEC]], %[[DBL_F]]
+// CIR:    cir.store %[[DBL_DEC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.double>, !cir.double
+// CIR:    cir.return %[[RV_LOAD]] : !cir.double
+
+// LLVM: define{{.*}} double @_Z13doubleUPreDecd({{.*}})
+// LLVM:   %[[DBL_F:.*]] = alloca double, align 8
+// LLVM:   %[[RV:.*]] = alloca double, align 8
+// LLVM:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_DEC:.*]] = fadd double %[[DBL_LOAD]], -1.000000e+00
+// LLVM:   store double %[[DBL_DEC]], ptr %[[DBL_F]], align 8
+// LLVM:   store double %[[DBL_DEC]], ptr %[[RV]], align 8
+// LLVM:   %[[RV_LOAD:.*]] = load double, ptr %[[RV]], align 8
+// LLVM:   ret double %[[RV_LOAD]]
+
+// OGCG: define{{.*}} double @_Z13doubleUPreDecd({{.*}})
+// OGCG:   %[[DBL_F:.*]] = alloca double, align 8
+// OGCG:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_DEC:.*]] = fadd double %[[DBL_LOAD]], -1.000000e+00
+// OGCG:   store double %[[DBL_DEC]], ptr %[[DBL_F]], align 8
+// OGCG:   ret double %[[DBL_DEC]]
+
+double doubleUPostInc(double f) {
+  return f++;
 }
 
-_Float16 fp16UPlus(_Float16 f) {
+// CIR: cir.func{{.*}} @_Z14doubleUPostIncd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+// CIR:    %[[DBL_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.double
+// CIR:    %[[DBL_INC:.*]] = cir.fadd %[[DBL_LOAD]], %[[DBL_ONE]]
+// CIR:    cir.store{{.*}} %[[DBL_INC]], %[[DBL_F]]
+// CIR:    cir.store %[[DBL_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.double>, !cir.double
+// CIR:    cir.return %[[RV_LOAD]] : !cir.double
+
+// LLVM: define{{.*}} double @_Z14doubleUPostIncd({{.*}})
+// LLVM:   %[[DBL_F:.*]] = alloca double, align 8
+// LLVM:   %[[RV:.*]] = alloca double, align 8
+// LLVM:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_INC:.*]] = fadd double %[[DBL_LOAD]], 1.000000e+00
+// LLVM:   store double %[[DBL_INC]], ptr %[[DBL_F]], align 8
+// LLVM:   store double %[[DBL_LOAD]], ptr %[[RV]], align 8
+// LLVM:   %[[RV_LOAD:.*]] = load double, ptr %[[RV]], align 8
+// LLVM:   ret double %[[RV_LOAD]]
+
+// OGCG: define{{.*}} double @_Z14doubleUPostIncd({{.*}})
+// OGCG:   %[[DBL_F:.*]] = alloca double, align 8
+// OGCG:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_INC:.*]] = fadd double %[[DBL_LOAD]], 1.000000e+00
+// OGCG:   store double %[[DBL_INC]], ptr %[[DBL_F]], align 8
+// OGCG:   ret double %[[DBL_LOAD]]
+
+double doubleUPostDec(double f) {
+  return f--;
+}
+
+// CIR: cir.func{{.*}} @_Z14doubleUPostDecd({{.*}}) -> (!cir.double{{.*}})
+// CIR:    %[[DBL_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.double>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.double>
+// CIR:    %[[DBL_LOAD:.*]] = cir.load{{.*}} %[[DBL_F]]
+// CIR:    %[[DBL_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.double
+// CIR:    %[[DBL_DEC:.*]] = cir.fadd %[[DBL_LOAD]], %[[DBL_NEGONE]]
+// CIR:    cir.store{{.*}} %[[DBL_DEC]], %[[DBL_F]]
+// CIR:    cir.store %[[DBL_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.double>, !cir.double
+// CIR:    cir.return %[[RV_LOAD]] : !cir.double
+
+// LLVM: define{{.*}} double @_Z14doubleUPostDecd({{.*}})
+// LLVM:   %[[DBL_F:.*]] = alloca double, align 8
+// LLVM:   %[[RV:.*]] = alloca double, align 8
+// LLVM:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// LLVM:   %[[DBL_DEC:.*]] = fadd double %[[DBL_LOAD]], -1.000000e+00
+// LLVM:   store double %[[DBL_DEC]], ptr %[[DBL_F]], align 8
+// LLVM:   store double %[[DBL_LOAD]], ptr %[[RV]], align 8
+// LLVM:   %[[RV_LOAD:.*]] = load double, ptr %[[RV]], align 8
+// LLVM:   ret double %[[RV_LOAD]]
+
+// OGCG: define{{.*}} double @_Z14doubleUPostDecd({{.*}})
+// OGCG:   %[[DBL_F:.*]] = alloca double, align 8
+// OGCG:   store double %{{.*}}, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_LOAD:.*]] = load double, ptr %[[DBL_F]], align 8
+// OGCG:   %[[DBL_DEC:.*]] = fadd double %[[DBL_LOAD]], -1.000000e+00
+// OGCG:   store double %[[DBL_DEC]], ptr %[[DBL_F]], align 8
+// OGCG:   ret double %[[DBL_LOAD]]
+
+// long double unary operations
+long double ldUPlus(long double f) {
   return +f;
 }
 
-// CHECK: cir.func{{.*}} @_Z9fp16UPlusDF16_({{.*}}) -> (!cir.f16{{.*}})
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[F:.*]]
-// CHECK:   %[[PROMOTED:.*]] = cir.cast floating %[[INPUT]] : !cir.f16 -> !cir.float
-// CHECK:   %[[UNPROMOTED:.*]] = cir.cast floating %[[PROMOTED]] : !cir.float -> !cir.f16
+// CIR: cir.func{{.*}} @_Z7ldUPluse({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+
+// LLVM: define{{.*}} x86_fp80 @_Z7ldUPluse({{.*}})
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %{{.*}}, align 16
+
+// OGCG: define{{.*}} x86_fp80 @_Z7ldUPluse({{.*}})
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %{{.*}}, align 16
+
+long double ldUMinus(long double f) {
+  return -f;
+}
+
+// CIR: cir.func{{.*}} @_Z8ldUMinuse({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+// CIR:    %[[LD_NEGATED:.*]] = cir.fneg %[[LD_LOAD]]
+
+// LLVM: define{{.*}} x86_fp80 @_Z8ldUMinuse({{.*}})
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %{{.*}}, align 16
+// LLVM:   %[[LD_NEGATED:.*]] = fneg x86_fp80 %[[LD_LOAD]]
+
+// OGCG: define{{.*}} x86_fp80 @_Z8ldUMinuse({{.*}})
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %{{.*}}, align 16
+// OGCG:   %[[LD_NEGATED:.*]] = fneg x86_fp80 %[[LD_LOAD]]
+
+long double ldUPreInc(long double f) {
+  return ++f;
+}
 
-// LLVM: define{{.*}} half @_Z9fp16UPlusDF16_({{.*}})
+// CIR: cir.func{{.*}} @_Z9ldUPreInce({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+// CIR:    %[[LD_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.long_double<!cir.f80>
+// CIR:    %[[LD_INC:.*]] = cir.fadd %[[LD_LOAD]], %[[LD_ONE]]
+// CIR:    cir.store{{.*}} %[[LD_INC]], %[[LD_F]]
+// CIR:    cir.store %[[LD_INC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.long_double<!cir.f80>>, !cir.long_double<!cir.f80>
+// CIR:    cir.return %[[RV_LOAD]] : !cir.long_double<!cir.f80>
+
+// LLVM: define{{.*}} x86_fp80 @_Z9ldUPreInce({{.*}})
+// LLVM:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// LLVM:   %[[RV:.*]] = alloca x86_fp80, align 16
+// LLVM:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_INC:.*]] = fadd x86_fp80 %[[LD_LOAD]], 1.000000e+00
+// LLVM:   store x86_fp80 %[[LD_INC]], ptr %[[LD_F]], align 16
+// LLVM:   store x86_fp80 %[[LD_INC]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load x86_fp80, ptr %[[RV]], align 16
+// LLVM:   ret x86_fp80 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} x86_fp80 @_Z9ldUPreInce({{.*}})
+// OGCG:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// OGCG:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_INC:.*]] = fadd x86_fp80 %[[LD_LOAD]], 1.000000e+00
+// OGCG:   store x86_fp80 %[[LD_INC]], ptr %[[LD_F]], align 16
+// OGCG:   ret x86_fp80 %[[LD_INC]]
+
+long double ldUPreDec(long double f) {
+  return --f;
+}
+
+// CIR: cir.func{{.*}} @_Z9ldUPreDece({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+// CIR:    %[[LD_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.long_double<!cir.f80>
+// CIR:    %[[LD_DEC:.*]] = cir.fadd %[[LD_LOAD]], %[[LD_NEGONE]]
+// CIR:    cir.store{{.*}} %[[LD_DEC]], %[[LD_F]]
+// CIR:    cir.store %[[LD_DEC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.long_double<!cir.f80>>, !cir.long_double<!cir.f80>
+// CIR:    cir.return %[[RV_LOAD]] : !cir.long_double<!cir.f80>
+
+// LLVM: define{{.*}} x86_fp80 @_Z9ldUPreDece({{.*}})
+// LLVM:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// LLVM:   %[[RV:.*]] = alloca x86_fp80, align 16
+// LLVM:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_DEC:.*]] = fadd x86_fp80 %[[LD_LOAD]], -1.000000e+00
+// LLVM:   store x86_fp80 %[[LD_DEC]], ptr %[[LD_F]], align 16
+// LLVM:   store x86_fp80 %[[LD_DEC]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load x86_fp80, ptr %[[RV]], align 16
+// LLVM:   ret x86_fp80 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} x86_fp80 @_Z9ldUPreDece({{.*}})
+// OGCG:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// OGCG:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_DEC:.*]] = fadd x86_fp80 %[[LD_LOAD]], -1.000000e+00
+// OGCG:   store x86_fp80 %[[LD_DEC]], ptr %[[LD_F]], align 16
+// OGCG:   ret x86_fp80 %[[LD_DEC]]
+
+long double ldUPostInc(long double f) {
+  return f++;
+}
+
+// CIR: cir.func{{.*}} @_Z10ldUPostInce({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+// CIR:    %[[LD_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.long_double<!cir.f80>
+// CIR:    %[[LD_INC:.*]] = cir.fadd %[[LD_LOAD]], %[[LD_ONE]]
+// CIR:    cir.store{{.*}} %[[LD_INC]], %[[LD_F]]
+// CIR:    cir.store %[[LD_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.long_double<!cir.f80>>, !cir.long_double<!cir.f80>
+// CIR:    cir.return %[[RV_LOAD]] : !cir.long_double<!cir.f80>
+
+// LLVM: define{{.*}} x86_fp80 @_Z10ldUPostInce({{.*}})
+// LLVM:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// LLVM:   %[[RV:.*]] = alloca x86_fp80, align 16
+// LLVM:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_INC:.*]] = fadd x86_fp80 %[[LD_LOAD]], 1.000000e+00
+// LLVM:   store x86_fp80 %[[LD_INC]], ptr %[[LD_F]], align 16
+// LLVM:   store x86_fp80 %[[LD_LOAD]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load x86_fp80, ptr %[[RV]], align 16
+// LLVM:   ret x86_fp80 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} x86_fp80 @_Z10ldUPostInce({{.*}})
+// OGCG:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// OGCG:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_INC:.*]] = fadd x86_fp80 %[[LD_LOAD]], 1.000000e+00
+// OGCG:   store x86_fp80 %[[LD_INC]], ptr %[[LD_F]], align 16
+// OGCG:   ret x86_fp80 %[[LD_LOAD]]
+
+long double ldUPostDec(long double f) {
+  return f--;
+}
+
+// CIR: cir.func{{.*}} @_Z10ldUPostDece({{.*}}) -> (!cir.long_double<!cir.f80>{{.*}})
+// CIR:    %[[LD_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.long_double<!cir.f80>>
+// CIR:    %[[LD_LOAD:.*]] = cir.load{{.*}} %[[LD_F]]
+// CIR:    %[[LD_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.long_double<!cir.f80>
+// CIR:    %[[LD_DEC:.*]] = cir.fadd %[[LD_LOAD]], %[[LD_NEGONE]]
+// CIR:    cir.store{{.*}} %[[LD_DEC]], %[[LD_F]]
+// CIR:    cir.store %[[LD_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.long_double<!cir.f80>>, !cir.long_double<!cir.f80>
+// CIR:    cir.return %[[RV_LOAD]] : !cir.long_double<!cir.f80>
+
+// LLVM: define{{.*}} x86_fp80 @_Z10ldUPostDece({{.*}})
+// LLVM:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// LLVM:   %[[RV:.*]] = alloca x86_fp80, align 16
+// LLVM:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// LLVM:   %[[LD_DEC:.*]] = fadd x86_fp80 %[[LD_LOAD]], -1.000000e+00
+// LLVM:   store x86_fp80 %[[LD_DEC]], ptr %[[LD_F]], align 16
+// LLVM:   store x86_fp80 %[[LD_LOAD]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load x86_fp80, ptr %[[RV]], align 16
+// LLVM:   ret x86_fp80 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} x86_fp80 @_Z10ldUPostDece({{.*}})
+// OGCG:   %[[LD_F:.*]] = alloca x86_fp80, align 16
+// OGCG:   store x86_fp80 %{{.*}}, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_LOAD:.*]] = load x86_fp80, ptr %[[LD_F]], align 16
+// OGCG:   %[[LD_DEC:.*]] = fadd x86_fp80 %[[LD_LOAD]], -1.000000e+00
+// OGCG:   store x86_fp80 %[[LD_DEC]], ptr %[[LD_F]], align 16
+// OGCG:   ret x86_fp80 %[[LD_LOAD]]
+
+// __float128 unary operations
+__float128 f128UPlus(__float128 f) {
+  return +f;
+}
+
+// CIR: cir.func{{.*}} @_Z9f128UPlusg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+
+// LLVM: define{{.*}} fp128 @_Z9f128UPlusg({{.*}})
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %{{.*}}, align 16
+
+// OGCG: define{{.*}} fp128 @_Z9f128UPlusg({{.*}})
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %{{.*}}, align 16
+
+__float128 f128UMinus(__float128 f) {
+  return -f;
+}
+
+// CIR: cir.func{{.*}} @_Z10f128UMinusg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+// CIR:    %[[F128_NEG:.*]] = cir.fneg %[[F128_LOAD]]
+
+// LLVM: define{{.*}} fp128 @_Z10f128UMinusg({{.*}})
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %{{.*}}, align 16
+// LLVM:   %[[F128_NEG:.*]] = fneg fp128 %[[F128_LOAD]]
+
+// OGCG: define{{.*}} fp128 @_Z10f128UMinusg({{.*}})
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %{{.*}}, align 16
+// OGCG:   %[[F128_NEG:.*]] = fneg fp128 %[[F128_LOAD]]
+
+__float128 f128UPreInc(__float128 f) {
+  return ++f;
+}
+
+// CIR: cir.func{{.*}} @_Z11f128UPreIncg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+// CIR:    %[[F128_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.f128
+// CIR:    %[[F128_INC:.*]] = cir.fadd %[[F128_LOAD]], %[[F128_ONE]]
+// CIR:    cir.store{{.*}} %[[F128_INC]], %[[F128_F]]
+// CIR:    cir.store %[[F128_INC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f128>, !cir.f128
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f128
+
+// LLVM: define{{.*}} fp128 @_Z11f128UPreIncg({{.*}})
+// LLVM:   %[[F128_F:.*]] = alloca fp128, align 16
+// LLVM:   %[[RV:.*]] = alloca fp128, align 16
+// LLVM:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_INC:.*]] = fadd fp128 %[[F128_LOAD]], 1.000000e+00
+// LLVM:   store fp128 %[[F128_INC]], ptr %[[F128_F]], align 16
+// LLVM:   store fp128 %[[F128_INC]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load fp128, ptr %[[RV]], align 16
+// LLVM:   ret fp128 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} fp128 @_Z11f128UPreIncg({{.*}})
+// OGCG:   %[[F128_F:.*]] = alloca fp128, align 16
+// OGCG:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_INC:.*]] = fadd fp128 %[[F128_LOAD]], 1.000000e+00
+// OGCG:   store fp128 %[[F128_INC]], ptr %[[F128_F]], align 16
+// OGCG:   ret fp128 %[[F128_INC]]
+
+__float128 f128UPreDec(__float128 f) {
+  return --f;
+}
+
+// CIR: cir.func{{.*}} @_Z11f128UPreDecg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+// CIR:    %[[F128_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.f128
+// CIR:    %[[F128_DEC:.*]] = cir.fadd %[[F128_LOAD]], %[[F128_NEGONE]]
+// CIR:    cir.store{{.*}} %[[F128_DEC]], %[[F128_F]]
+// CIR:    cir.store %[[F128_DEC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f128>, !cir.f128
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f128
+
+// LLVM: define{{.*}} fp128 @_Z11f128UPreDecg({{.*}})
+// LLVM:   %[[F128_F:.*]] = alloca fp128, align 16
+// LLVM:   %[[RV:.*]] = alloca fp128, align 16
+// LLVM:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_DEC:.*]] = fadd fp128 %[[F128_LOAD]], -1.000000e+00
+// LLVM:   store fp128 %[[F128_DEC]], ptr %[[F128_F]], align 16
+// LLVM:   store fp128 %[[F128_DEC]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load fp128, ptr %[[RV]], align 16
+// LLVM:   ret fp128 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} fp128 @_Z11f128UPreDecg({{.*}})
+// OGCG:   %[[F128_F:.*]] = alloca fp128, align 16
+// OGCG:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_DEC:.*]] = fadd fp128 %[[F128_LOAD]], -1.000000e+00
+// OGCG:   store fp128 %[[F128_DEC]], ptr %[[F128_F]], align 16
+// OGCG:   ret fp128 %[[F128_DEC]]
+
+__float128 f128UPostInc(__float128 f) {
+  return f++;
+}
+
+// CIR: cir.func{{.*}} @_Z12f128UPostIncg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+// CIR:    %[[F128_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.f128
+// CIR:    %[[F128_INC:.*]] = cir.fadd %[[F128_LOAD]], %[[F128_ONE]]
+// CIR:    cir.store{{.*}} %[[F128_INC]], %[[F128_F]]
+// CIR:    cir.store %[[F128_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f128>, !cir.f128
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f128
+
+// LLVM: define{{.*}} fp128 @_Z12f128UPostIncg({{.*}})
+// LLVM:   %[[F128_F:.*]] = alloca fp128, align 16
+// LLVM:   %[[RV:.*]] = alloca fp128, align 16
+// LLVM:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_INC:.*]] = fadd fp128 %[[F128_LOAD]], 1.000000e+00
+// LLVM:   store fp128 %[[F128_INC]], ptr %[[F128_F]], align 16
+// LLVM:   store fp128 %[[F128_LOAD]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load fp128, ptr %[[RV]], align 16
+// LLVM:   ret fp128 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} fp128 @_Z12f128UPostIncg({{.*}})
+// OGCG:   %[[F128_F:.*]] = alloca fp128, align 16
+// OGCG:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_INC:.*]] = fadd fp128 %[[F128_LOAD]], 1.000000e+00
+// OGCG:   store fp128 %[[F128_INC]], ptr %[[F128_F]], align 16
+// OGCG:   ret fp128 %[[F128_LOAD]]
+
+__float128 f128UPostDec(__float128 f) {
+  return f--;
+}
+
+// CIR: cir.func{{.*}} @_Z12f128UPostDecg({{.*}}) -> (!cir.f128{{.*}})
+// CIR:    %[[F128_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f128>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f128>
+// CIR:    %[[F128_LOAD:.*]] = cir.load{{.*}} %[[F128_F]]
+// CIR:    %[[F128_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.f128
+// CIR:    %[[F128_DEC:.*]] = cir.fadd %[[F128_LOAD]], %[[F128_NEGONE]]
+// CIR:    cir.store{{.*}} %[[F128_DEC]], %[[F128_F]]
+// CIR:    cir.store %[[F128_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f128>, !cir.f128
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f128
+
+// LLVM: define{{.*}} fp128 @_Z12f128UPostDecg({{.*}})
+// LLVM:   %[[F128_F:.*]] = alloca fp128, align 16
+// LLVM:   %[[RV:.*]] = alloca fp128, align 16
+// LLVM:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// LLVM:   %[[F128_DEC:.*]] = fadd fp128 %[[F128_LOAD]], -1.000000e+00
+// LLVM:   store fp128 %[[F128_DEC]], ptr %[[F128_F]], align 16
+// LLVM:   store fp128 %[[F128_LOAD]], ptr %[[RV]], align 16
+// LLVM:   %[[RV_LOAD:.*]] = load fp128, ptr %[[RV]], align 16
+// LLVM:   ret fp128 %[[RV_LOAD]]
+
+// OGCG: define{{.*}} fp128 @_Z12f128UPostDecg({{.*}})
+// OGCG:   %[[F128_F:.*]] = alloca fp128, align 16
+// OGCG:   store fp128 %{{.*}}, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_LOAD:.*]] = load fp128, ptr %[[F128_F]], align 16
+// OGCG:   %[[F128_DEC:.*]] = fadd fp128 %[[F128_LOAD]], -1.000000e+00
+// OGCG:   store fp128 %[[F128_DEC]], ptr %[[F128_F]], align 16
+// OGCG:   ret fp128 %[[F128_LOAD]]
+
+// Float16 unary operations
+_Float16 Float16UPlus(_Float16 f) {
+  return +f;
+}
+
+// CIR: cir.func{{.*}} @_Z12Float16UPlusDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[F:.*]]
+// CIR:    %[[PROMOTED:.*]] = cir.cast floating %[[INPUT]] : !cir.f16 -> !cir.float
+// CIR:    %[[UNPROMOTED:.*]] = cir.cast floating %[[PROMOTED]] : !cir.float -> !cir.f16
+
+// LLVM: define{{.*}} half @_Z12Float16UPlusDF16_({{.*}})
 // LLVM:   %[[F_LOAD:.*]] = load half, ptr %{{.*}}, align 2
 // LLVM:   %[[PROMOTED:.*]] = fpext half %[[F_LOAD]] to float
 // LLVM:   %[[UNPROMOTED:.*]] = fptrunc float %[[PROMOTED]] to half
 
-// OGCG: define{{.*}} half @_Z9fp16UPlusDF16_({{.*}})
+// OGCG: define{{.*}} half @_Z12Float16UPlusDF16_({{.*}})
 // OGCG:   %[[F_LOAD:.*]] = load half, ptr %{{.*}}, align 2
 // OGCG:   %[[PROMOTED:.*]] = fpext half %[[F_LOAD]] to float
 // OGCG:   %[[UNPROMOTED:.*]] = fptrunc float %[[PROMOTED]] to half
 
-_Float16 fp16UMinus(_Float16 f) {
+_Float16 Float16UMinus(_Float16 f) {
   return -f;
 }
 
-// CHECK: cir.func{{.*}} @_Z10fp16UMinusDF16_({{.*}}) -> (!cir.f16{{.*}})
-// CHECK:   %[[INPUT:.*]] = cir.load{{.*}} %[[F:.*]]
-// CHECK:   %[[PROMOTED:.*]] = cir.cast floating %[[INPUT]] : !cir.f16 -> !cir.float
-// CHECK:   %[[RESULT:.*]] = cir.fneg %[[PROMOTED]]
-// CHECK:   %[[UNPROMOTED:.*]] = cir.cast floating %[[RESULT]] : !cir.float -> !cir.f16
+// CIR: cir.func{{.*}} @_Z13Float16UMinusDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[INPUT:.*]] = cir.load{{.*}} %[[F:.*]]
+// CIR:    %[[PROMOTED:.*]] = cir.cast floating %[[INPUT]] : !cir.f16 -> !cir.float
+// CIR:    %[[RESULT:.*]] = cir.fneg %[[PROMOTED]]
+// CIR:    %[[UNPROMOTED:.*]] = cir.cast floating %[[RESULT]] : !cir.float -> !cir.f16
 
-// LLVM: define{{.*}} half @_Z10fp16UMinusDF16_({{.*}})
+// LLVM: define{{.*}} half @_Z13Float16UMinusDF16_({{.*}})
 // LLVM:   %[[F_LOAD:.*]] = load half, ptr %{{.*}}, align 2
 // LLVM:   %[[PROMOTED:.*]] = fpext half %[[F_LOAD]] to float
 // LLVM:   %[[RESULT:.*]] = fneg float %[[PROMOTED]]
 // LLVM:   %[[UNPROMOTED:.*]] = fptrunc float %[[RESULT]] to half
 
-// OGCG: define{{.*}} half @_Z10fp16UMinusDF16_({{.*}})
+// OGCG: define{{.*}} half @_Z13Float16UMinusDF16_({{.*}})
 // OGCG:   %[[F_LOAD:.*]] = load half, ptr %{{.*}}, align 2
 // OGCG:   %[[PROMOTED:.*]] = fpext half %[[F_LOAD]] to float
 // OGCG:   %[[RESULT:.*]] = fneg float %[[PROMOTED]]
 // OGCG:   %[[UNPROMOTED:.*]] = fptrunc float %[[RESULT]] to half
 
+_Float16 Float16UPreInc(_Float16 f) {
+  return ++f;
+}
+
+// CIR: cir.func{{.*}} @_Z14Float16UPreIncDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[PREINC_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:    %[[PREINC_INPUT:.*]] = cir.load{{.*}} %[[PREINC_F]]
+// CIR:    %[[PREINC_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.f16
+// CIR:    %[[PREINC_RESULT:.*]] = cir.fadd %[[PREINC_INPUT]], %[[PREINC_ONE]] : !cir.f16
+// CIR:    cir.store{{.*}} %[[PREINC_RESULT]], %[[PREINC_F]]
+// CIR:    cir.store %[[PREINC_RESULT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f16
+
+// LLVM: define{{.*}} half @_Z14Float16UPreIncDF16_({{.*}})
+// LLVM:   %[[PREINC_F_ADDR:.*]] = alloca half, align 2
+// LLVM:   %[[RV:.*]] = alloca half, align 2
+// LLVM:   store half %{{.*}}, ptr %[[PREINC_F_ADDR]], align 2
+// LLVM:   %[[PREINC_INPUT:.*]] = load half, ptr %[[PREINC_F_ADDR]], align 2
+// LLVM:   %[[PREINC_RESULT:.*]] = fadd half %[[PREINC_INPUT]], 1.000000e+00
+// LLVM:   store half %[[PREINC_RESULT]], ptr %[[PREINC_F_ADDR]], align 2
+// LLVM:   store half %[[PREINC_RESULT]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load half, ptr %[[RV]], align 2
+// LLVM:   ret half %[[RV_LOAD]]
+
+// OGCG: define{{.*}} half @_Z14Float16UPreIncDF16_({{.*}})
+// OGCG:   %[[PREINC_F_ADDR:.*]] = alloca half, align 2
+// OGCG:   store half %{{.*}}, ptr %[[PREINC_F_ADDR]], align 2
+// OGCG:   %[[PREINC_INPUT:.*]] = load half, ptr %[[PREINC_F_ADDR]], align 2
+// OGCG:   %[[PREINC_RESULT:.*]] = fadd half %[[PREINC_INPUT]], 1.000000e+00
+// OGCG:   store half %[[PREINC_RESULT]], ptr %[[PREINC_F_ADDR]], align 2
+// OGCG:   ret half %[[PREINC_RESULT]]
+
+_Float16 Float16UPreDec(_Float16 f) {
+  return --f;
+}
+
+// CIR: cir.func{{.*}} @_Z14Float16UPreDecDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[PREDEC_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:    %[[PREDEC_INPUT:.*]] = cir.load{{.*}} %[[PREDEC_F]]
+// CIR:    %[[PREDEC_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.f16
+// CIR:    %[[PREDEC_RESULT:.*]] = cir.fadd %[[PREDEC_INPUT]], %[[PREDEC_NEGONE]] : !cir.f16
+// CIR:    cir.store{{.*}} %[[PREDEC_RESULT]], %[[PREDEC_F]]
+// CIR:    cir.store %[[PREDEC_RESULT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f16
+
+// LLVM: define{{.*}} half @_Z14Float16UPreDecDF16_({{.*}})
+// LLVM:   %[[PREDEC_F_ADDR:.*]] = alloca half, align 2
+// LLVM:   %[[RV:.*]] = alloca half, align 2
+// LLVM:   store half %{{.*}}, ptr %[[PREDEC_F_ADDR]], align 2
+// LLVM:   %[[PREDEC_INPUT:.*]] = load half, ptr %[[PREDEC_F_ADDR]], align 2
+// LLVM:   %[[PREDEC_RESULT:.*]] = fadd half %[[PREDEC_INPUT]], -1.000000e+00
+// LLVM:   store half %[[PREDEC_RESULT]], ptr %[[PREDEC_F_ADDR]], align 2
+// LLVM:   store half %[[PREDEC_RESULT]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load half, ptr %[[RV]], align 2
+// LLVM:   ret half %[[RV_LOAD]]
+
+// OGCG: define{{.*}} half @_Z14Float16UPreDecDF16_({{.*}})
+// OGCG:   %[[PREDEC_F_ADDR:.*]] = alloca half, align 2
+// OGCG:   store half %{{.*}}, ptr %[[PREDEC_F_ADDR]], align 2
+// OGCG:   %[[PREDEC_INPUT:.*]] = load half, ptr %[[PREDEC_F_ADDR]], align 2
+// OGCG:   %[[PREDEC_RESULT:.*]] = fadd half %[[PREDEC_INPUT]], -1.000000e+00
+// OGCG:   store half %[[PREDEC_RESULT]], ptr %[[PREDEC_F_ADDR]], align 2
+// OGCG:   ret half %[[PREDEC_RESULT]]
+
+_Float16 Float16UPostInc(_Float16 f) {
+  return f++;
+}
+
+// CIR: cir.func{{.*}} @_Z15Float16UPostIncDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[POSTINC_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:    %[[POSTINC_INPUT:.*]] = cir.load{{.*}} %[[POSTINC_F]]
+// CIR:    %[[POSTINC_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.f16
+// CIR:    %[[POSTINC_RESULT:.*]] = cir.fadd %[[POSTINC_INPUT]], %[[POSTINC_ONE]] : !cir.f16
+// CIR:    cir.store{{.*}} %[[POSTINC_RESULT]], %[[POSTINC_F]]
+// CIR:    cir.store %[[POSTINC_INPUT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f16
+
+// LLVM: define{{.*}} half @_Z15Float16UPostIncDF16_({{.*}})
+// LLVM:   %[[POSTINC_F_ADDR:.*]] = alloca half, align 2
+// LLVM:   %[[RV:.*]] = alloca half, align 2
+// LLVM:   store half %{{.*}}, ptr %[[POSTINC_F_ADDR]], align 2
+// LLVM:   %[[POSTINC_INPUT:.*]] = load half, ptr %[[POSTINC_F_ADDR]], align 2
+// LLVM:   %[[POSTINC_RESULT:.*]] = fadd half %[[POSTINC_INPUT]], 1.000000e+00
+// LLVM:   store half %[[POSTINC_RESULT]], ptr %[[POSTINC_F_ADDR]], align 2
+// LLVM:   store half %[[POSTINC_INPUT]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load half, ptr %[[RV]], align 2
+// LLVM:   ret half %[[RV_LOAD]]
+
+// OGCG: define{{.*}} half @_Z15Float16UPostIncDF16_({{.*}})
+// OGCG:   %[[POSTINC_F_ADDR:.*]] = alloca half, align 2
+// OGCG:   store half %{{.*}}, ptr %[[POSTINC_F_ADDR]], align 2
+// OGCG:   %[[POSTINC_INPUT:.*]] = load half, ptr %[[POSTINC_F_ADDR]], align 2
+// OGCG:   %[[POSTINC_RESULT:.*]] = fadd half %[[POSTINC_INPUT]], 1.000000e+00
+// OGCG:   store half %[[POSTINC_RESULT]], ptr %[[POSTINC_F_ADDR]], align 2
+// OGCG:   ret half %[[POSTINC_INPUT]]
+
+_Float16 Float16UPostDec(_Float16 f) {
+  return f--;
+}
+
+// CIR: cir.func{{.*}} @_Z15Float16UPostDecDF16_({{.*}}) -> (!cir.f16{{.*}})
+// CIR:    %[[POSTDEC_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:    %[[POSTDEC_INPUT:.*]] = cir.load{{.*}} %[[POSTDEC_F]]
+// CIR:    %[[POSTDEC_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.f16
+// CIR:    %[[POSTDEC_RESULT:.*]] = cir.fadd %[[POSTDEC_INPUT]], %[[POSTDEC_NEGONE]] : !cir.f16
+// CIR:    cir.store{{.*}} %[[POSTDEC_RESULT]], %[[POSTDEC_F]]
+// CIR:    cir.store %[[POSTDEC_INPUT]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.f16
+
+// LLVM: define{{.*}} half @_Z15Float16UPostDecDF16_({{.*}})
+// LLVM:   %[[POSTDEC_F_ADDR:.*]] = alloca half, align 2
+// LLVM:   %[[RV:.*]] = alloca half, align 2
+// LLVM:   store half %{{.*}}, ptr %[[POSTDEC_F_ADDR]], align 2
+// LLVM:   %[[POSTDEC_INPUT:.*]] = load half, ptr %[[POSTDEC_F_ADDR]], align 2
+// LLVM:   %[[POSTDEC_RESULT:.*]] = fadd half %[[POSTDEC_INPUT]], -1.000000e+00
+// LLVM:   store half %[[POSTDEC_RESULT]], ptr %[[POSTDEC_F_ADDR]], align 2
+// LLVM:   store half %[[POSTDEC_INPUT]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load half, ptr %[[RV]], align 2
+// LLVM:   ret half %[[RV_LOAD]]
+
+// OGCG: define{{.*}} half @_Z15Float16UPostDecDF16_({{.*}})
+// OGCG:   %[[POSTDEC_F_ADDR:.*]] = alloca half, align 2
+// OGCG:   store half %{{.*}}, ptr %[[POSTDEC_F_ADDR]], align 2
+// OGCG:   %[[POSTDEC_INPUT:.*]] = load half, ptr %[[POSTDEC_F_ADDR]], align 2
+// OGCG:   %[[POSTDEC_RESULT:.*]] = fadd half %[[POSTDEC_INPUT]], -1.000000e+00
+// OGCG:   store half %[[POSTDEC_RESULT]], ptr %[[POSTDEC_F_ADDR]], align 2
+// OGCG:   ret half %[[POSTDEC_INPUT]]
+
+// __fp16 unary operations
+void fp16PtrUPlus(__fp16 *f) {
+  *f = +(*f);
+}
+
+// CIR: cir.func{{.*}} @_Z12fp16PtrUPlusPDh({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_PROMOTED]] : !cir.float -> !cir.f16
+
+// LLVM: define{{.*}} void @_Z12fp16PtrUPlusPDh({{.*}})
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %{{.*}}, align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_PROMOTED]] to half
+
+// OGCG: define{{.*}} void @_Z12fp16PtrUPlusPDh({{.*}})
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %{{.*}}, align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_PROMOTED]] to half
+
+void fp16PtrUMinus(__fp16 *f) {
+  *f = -(*f);
+}
+
+// CIR: cir.func{{.*}} @_Z13fp16PtrUMinusPDh({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_NEGATED:.*]] = cir.fneg %[[FPTR_PROMOTED]]
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_NEGATED]] : !cir.float -> !cir.f16
+
+// LLVM: define{{.*}} void @_Z13fp16PtrUMinusPDh({{.*}})
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %{{.*}}, align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_NEGATED:.*]] = fneg float %[[FPTR_PROMOTED]]
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_NEGATED]] to half
+
+// OGCG: define{{.*}} void @_Z13fp16PtrUMinusPDh({{.*}})
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %{{.*}}, align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_NEGATED:.*]] = fneg float %[[FPTR_PROMOTED]]
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_NEGATED]] to half
+
+void fp16PtrUPreInc(__fp16 *f, __fp16 *ret) {
+  *ret = ++(*f);
+}
+
+// CIR: cir.func{{.*}} @_Z14fp16PtrUPreIncPDhS_({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_RET:.*]] = cir.alloca "ret" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    %[[FPTR_INC:.*]] = cir.fadd %[[FPTR_PROMOTED]], %[[FPTR_ONE]] : !cir.float
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_INC]] : !cir.float -> !cir.f16
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[FPTR_DEREF]]
+// CIR:    %[[RET_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_RET]]
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[RET_DEREF]]
+
+// LLVM: define{{.*}} void @_Z14fp16PtrUPreIncPDhS_({{.*}})
+// LLVM:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// LLVM:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_INC:.*]] = fadd float %[[FPTR_PROMOTED]], 1.000000e+00
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_INC]] to half
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// LLVM:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[RET_ADDR]], align 2
+
+// OGCG: define{{.*}} void @_Z14fp16PtrUPreIncPDhS_({{.*}})
+// OGCG:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// OGCG:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_INC:.*]] = fadd float %[[FPTR_PROMOTED]], 1.000000e+00
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_INC]] to half
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// OGCG:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[RET_ADDR]], align 2
+
+void fp16PtrUPreDec(__fp16 *f, __fp16 *ret) {
+  *ret = --(*f);
+}
+
+// CIR: cir.func{{.*}} @_Z14fp16PtrUPreDecPDhS_({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_RET:.*]] = cir.alloca "ret" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:    %[[FPTR_DEC:.*]] = cir.fadd %[[FPTR_PROMOTED]], %[[FPTR_NEGONE]] : !cir.float
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_DEC]] : !cir.float -> !cir.f16
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[FPTR_DEREF]]
+// CIR:    %[[RET_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_RET]]
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[RET_DEREF]]
+
+// LLVM: define{{.*}} void @_Z14fp16PtrUPreDecPDhS_({{.*}})
+// LLVM:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// LLVM:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_DEC:.*]] = fadd float %[[FPTR_PROMOTED]], -1.000000e+00
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_DEC]] to half
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// LLVM:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[RET_ADDR]], align 2
+
+// OGCG: define{{.*}} void @_Z14fp16PtrUPreDecPDhS_({{.*}})
+// OGCG:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// OGCG:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_DEC:.*]] = fadd float %[[FPTR_PROMOTED]], -1.000000e+00
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_DEC]] to half
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// OGCG:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[RET_ADDR]], align 2
+
+void fp16PtrUPostInc(__fp16 *f, __fp16 *ret) {
+  *ret = (*f)++;
+}
+
+// CIR: cir.func{{.*}} @_Z15fp16PtrUPostIncPDhS_({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_RET:.*]] = cir.alloca "ret" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:    %[[FPTR_INC:.*]] = cir.fadd %[[FPTR_PROMOTED]], %[[FPTR_ONE]] : !cir.float
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_INC]] : !cir.float -> !cir.f16
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[FPTR_DEREF]]
+// CIR:    %[[RET_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_RET]]
+// CIR:    cir.store{{.*}} %[[FPTR_LOAD]], %[[RET_DEREF]]
+
+// LLVM: define{{.*}} void @_Z15fp16PtrUPostIncPDhS_({{.*}})
+// LLVM:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// LLVM:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_INC:.*]] = fadd float %[[FPTR_PROMOTED]], 1.000000e+00
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_INC]] to half
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// LLVM:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   store half %[[FPTR_LOAD]], ptr %[[RET_ADDR]], align 2
+
+// OGCG: define{{.*}} void @_Z15fp16PtrUPostIncPDhS_({{.*}})
+// OGCG:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// OGCG:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_INC:.*]] = fadd float %[[FPTR_PROMOTED]], 1.000000e+00
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_INC]] to half
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// OGCG:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   store half %[[FPTR_LOAD]], ptr %[[RET_ADDR]], align 2
+
+void fp16PtrUPostDec(__fp16 *f, __fp16 *ret) {
+  *ret = (*f)--;
+}
+
+// CIR: cir.func{{.*}} @_Z15fp16PtrUPostDecPDhS_({{.*}})
+// CIR:    %[[FPTR_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_RET:.*]] = cir.alloca "ret" {{.*}} init : !cir.ptr<!cir.ptr<!cir.f16>>
+// CIR:    %[[FPTR_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_F]]
+// CIR:    %[[FPTR_LOAD:.*]] = cir.load{{.*}} %[[FPTR_DEREF]]
+// CIR:    %[[FPTR_PROMOTED:.*]] = cir.cast floating %[[FPTR_LOAD]] : !cir.f16 -> !cir.float
+// CIR:    %[[FPTR_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:    %[[FPTR_DEC:.*]] = cir.fadd %[[FPTR_PROMOTED]], %[[FPTR_NEGONE]] : !cir.float
+// CIR:    %[[FPTR_UNPROMOTED:.*]] = cir.cast floating %[[FPTR_DEC]] : !cir.float -> !cir.f16
+// CIR:    cir.store{{.*}} %[[FPTR_UNPROMOTED]], %[[FPTR_DEREF]]
+// CIR:    %[[RET_DEREF:.*]] = cir.load deref{{.*}} %[[FPTR_RET]]
+// CIR:    cir.store{{.*}} %[[FPTR_LOAD]], %[[RET_DEREF]]
+
+// LLVM: define{{.*}} void @_Z15fp16PtrUPostDecPDhS_({{.*}})
+// LLVM:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// LLVM:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// LLVM:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// LLVM:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// LLVM:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// LLVM:   %[[FPTR_DEC:.*]] = fadd float %[[FPTR_PROMOTED]], -1.000000e+00
+// LLVM:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_DEC]] to half
+// LLVM:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// LLVM:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// LLVM:   store half %[[FPTR_LOAD]], ptr %[[RET_ADDR]], align 2
+
+// OGCG: define{{.*}} void @_Z15fp16PtrUPostDecPDhS_({{.*}})
+// OGCG:   %[[F_ADDR:.*]] = alloca ptr, align 8
+// OGCG:   %[[RET_ADDR_PTR:.*]] = alloca ptr, align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[F_ADDR]], align 8
+// OGCG:   store ptr %{{.*}}, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   %[[F_PTR:.*]] = load ptr, ptr %[[F_ADDR]], align 8
+// OGCG:   %[[FPTR_LOAD:.*]] = load half, ptr %[[F_PTR]], align 2
+// OGCG:   %[[FPTR_PROMOTED:.*]] = fpext half %[[FPTR_LOAD]] to float
+// OGCG:   %[[FPTR_DEC:.*]] = fadd float %[[FPTR_PROMOTED]], -1.000000e+00
+// OGCG:   %[[FPTR_UNPROMOTED:.*]] = fptrunc float %[[FPTR_DEC]] to half
+// OGCG:   store half %[[FPTR_UNPROMOTED]], ptr %[[F_PTR]], align 2
+// OGCG:   %[[RET_ADDR:.*]] = load ptr, ptr %[[RET_ADDR_PTR]], align 8
+// OGCG:   store half %[[FPTR_LOAD]], ptr %[[RET_ADDR]], align 2
+
+// __bf16 unary operations
+__bf16 bf16UPlus(__bf16 f) {
+  return +f;
+}
+
+// CIR: cir.func{{.*}} @_Z9bf16UPlusDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_PROMOTED:.*]] = cir.cast floating %[[BF16_LOAD]] : !cir.bf16 -> !cir.float
+// CIR:    %[[BF16_UNPROMOTED:.*]] = cir.cast floating %[[BF16_PROMOTED]] : !cir.float -> !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z9bf16UPlusDF16b({{.*}})
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %{{.*}}, align 2
+// LLVM:   %[[BF16_PROMOTED:.*]] = fpext bfloat %[[BF16_LOAD]] to float
+// LLVM:   %[[BF16_UNPROMOTED:.*]] = fptrunc float %[[BF16_PROMOTED]] to bfloat
+
+// OGCG: define{{.*}} bfloat @_Z9bf16UPlusDF16b({{.*}})
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %{{.*}}, align 2
+// OGCG:   %[[BF16_PROMOTED:.*]] = fpext bfloat %[[BF16_LOAD]] to float
+// OGCG:   %[[BF16_UNPROMOTED:.*]] = fptrunc float %[[BF16_PROMOTED]] to bfloat
+
+__bf16 bf16UMinus(__bf16 f) {
+  return -f;
+}
+
+// CIR: cir.func{{.*}} @_Z10bf16UMinusDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_PROMOTED:.*]] = cir.cast floating %[[BF16_LOAD]] : !cir.bf16 -> !cir.float
+// CIR:    %[[BF16_NEGATED:.*]] = cir.fneg %[[BF16_PROMOTED]]
+// CIR:    %[[BF16_UNPROMOTED:.*]] = cir.cast floating %[[BF16_NEGATED]] : !cir.float -> !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z10bf16UMinusDF16b({{.*}})
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %{{.*}}, align 2
+// LLVM:   %[[BF16_PROMOTED:.*]] = fpext bfloat %[[BF16_LOAD]] to float
+// LLVM:   %[[BF16_NEGATED:.*]] = fneg float %[[BF16_PROMOTED]]
+// LLVM:   %[[BF16_UNPROMOTED:.*]] = fptrunc float %[[BF16_NEGATED]] to bfloat
+
+// OGCG: define{{.*}} bfloat @_Z10bf16UMinusDF16b({{.*}})
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %{{.*}}, align 2
+// OGCG:   %[[BF16_PROMOTED:.*]] = fpext bfloat %[[BF16_LOAD]] to float
+// OGCG:   %[[BF16_NEGATED:.*]] = fneg float %[[BF16_PROMOTED]]
+// OGCG:   %[[BF16_UNPROMOTED:.*]] = fptrunc float %[[BF16_NEGATED]] to bfloat
+
+__bf16 bf16UPreInc(__bf16 f) {
+  return ++f;
+}
+
+// CIR: cir.func{{.*}} @_Z11bf16UPreIncDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.bf16
+// CIR:    %[[BF16_INC:.*]] = cir.fadd %[[BF16_LOAD]], %[[BF16_ONE]] : !cir.bf16
+// CIR:    cir.store{{.*}} %[[BF16_INC]], %[[BF16_F]]
+// CIR:    cir.store %[[BF16_INC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.bf16>, !cir.bf16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z11bf16UPreIncDF16b({{.*}})
+// LLVM:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// LLVM:   %[[RV:.*]] = alloca bfloat, align 2
+// LLVM:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_INC:.*]] = fadd bfloat %[[BF16_LOAD]], 1.000000e+00
+// LLVM:   store bfloat %[[BF16_INC]], ptr %[[BF16_F]], align 2
+// LLVM:   store bfloat %[[BF16_INC]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load bfloat, ptr %[[RV]], align 2
+// LLVM:   ret bfloat %[[RV_LOAD]]
+
+// OGCG: define{{.*}} bfloat @_Z11bf16UPreIncDF16b({{.*}})
+// OGCG:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// OGCG:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_INC:.*]] = fadd bfloat %[[BF16_LOAD]], 1.000000e+00
+// OGCG:   store bfloat %[[BF16_INC]], ptr %[[BF16_F]], align 2
+// OGCG:   ret bfloat %[[BF16_INC]]
+
+__bf16 bf16UPreDec(__bf16 f) {
+  return --f;
+}
+
+// CIR: cir.func{{.*}} @_Z11bf16UPreDecDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.bf16
+// CIR:    %[[BF16_DEC:.*]] = cir.fadd %[[BF16_LOAD]], %[[BF16_NEGONE]] : !cir.bf16
+// CIR:    cir.store{{.*}} %[[BF16_DEC]], %[[BF16_F]]
+// CIR:    cir.store %[[BF16_DEC]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.bf16>, !cir.bf16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z11bf16UPreDecDF16b({{.*}})
+// LLVM:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// LLVM:   %[[RV:.*]] = alloca bfloat, align 2
+// LLVM:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_DEC:.*]] = fadd bfloat %[[BF16_LOAD]], -1.000000e+00
+// LLVM:   store bfloat %[[BF16_DEC]], ptr %[[BF16_F]], align 2
+// LLVM:   store bfloat %[[BF16_DEC]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load bfloat, ptr %[[RV]], align 2
+// LLVM:   ret bfloat %[[RV_LOAD]]
+
+// OGCG: define{{.*}} bfloat @_Z11bf16UPreDecDF16b({{.*}})
+// OGCG:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// OGCG:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_DEC:.*]] = fadd bfloat %[[BF16_LOAD]], -1.000000e+00
+// OGCG:   store bfloat %[[BF16_DEC]], ptr %[[BF16_F]], align 2
+// OGCG:   ret bfloat %[[BF16_DEC]]
+
+__bf16 bf16UPostInc(__bf16 f) {
+  return f++;
+}
+
+// CIR: cir.func{{.*}} @_Z12bf16UPostIncDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.bf16
+// CIR:    %[[BF16_INC:.*]] = cir.fadd %[[BF16_LOAD]], %[[BF16_ONE]] : !cir.bf16
+// CIR:    cir.store{{.*}} %[[BF16_INC]], %[[BF16_F]]
+// CIR:    cir.store %[[BF16_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.bf16>, !cir.bf16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z12bf16UPostIncDF16b({{.*}})
+// LLVM:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// LLVM:   %[[RV:.*]] = alloca bfloat, align 2
+// LLVM:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_INC:.*]] = fadd bfloat %[[BF16_LOAD]], 1.000000e+00
+// LLVM:   store bfloat %[[BF16_INC]], ptr %[[BF16_F]], align 2
+// LLVM:   store bfloat %[[BF16_LOAD]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load bfloat, ptr %[[RV]], align 2
+// LLVM:   ret bfloat %[[RV_LOAD]]
+
+// OGCG: define{{.*}} bfloat @_Z12bf16UPostIncDF16b({{.*}})
+// OGCG:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// OGCG:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_INC:.*]] = fadd bfloat %[[BF16_LOAD]], 1.000000e+00
+// OGCG:   store bfloat %[[BF16_INC]], ptr %[[BF16_F]], align 2
+// OGCG:   ret bfloat %[[BF16_LOAD]]
+
+__bf16 bf16UPostDec(__bf16 f) {
+  return f--;
+}
+
+// CIR: cir.func{{.*}} @_Z12bf16UPostDecDF16b({{.*}}) -> (!cir.bf16{{.*}})
+// CIR:    %[[BF16_F:.*]] = cir.alloca "f" {{.*}} init : !cir.ptr<!cir.bf16>
+// CIR:    %[[RV:.*]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.bf16>
+// CIR:    %[[BF16_LOAD:.*]] = cir.load{{.*}} %[[BF16_F]]
+// CIR:    %[[BF16_NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.bf16
+// CIR:    %[[BF16_DEC:.*]] = cir.fadd %[[BF16_LOAD]], %[[BF16_NEGONE]] : !cir.bf16
+// CIR:    cir.store{{.*}} %[[BF16_DEC]], %[[BF16_F]]
+// CIR:    cir.store %[[BF16_LOAD]], %[[RV]]
+// CIR:    %[[RV_LOAD:.*]] = cir.load %[[RV]] : !cir.ptr<!cir.bf16>, !cir.bf16
+// CIR:    cir.return %[[RV_LOAD]] : !cir.bf16
+
+// LLVM: define{{.*}} bfloat @_Z12bf16UPostDecDF16b({{.*}})
+// LLVM:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// LLVM:   %[[RV:.*]] = alloca bfloat, align 2
+// LLVM:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// LLVM:   %[[BF16_DEC:.*]] = fadd bfloat %[[BF16_LOAD]], -1.000000e+00
+// LLVM:   store bfloat %[[BF16_DEC]], ptr %[[BF16_F]], align 2
+// LLVM:   store bfloat %[[BF16_LOAD]], ptr %[[RV]], align 2
+// LLVM:   %[[RV_LOAD:.*]] = load bfloat, ptr %[[RV]], align 2
+// LLVM:   ret bfloat %[[RV_LOAD]]
+
+// OGCG: define{{.*}} bfloat @_Z12bf16UPostDecDF16b({{.*}})
+// OGCG:   %[[BF16_F:.*]] = alloca bfloat, align 2
+// OGCG:   store bfloat %{{.*}}, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_LOAD:.*]] = load bfloat, ptr %[[BF16_F]], align 2
+// OGCG:   %[[BF16_DEC:.*]] = fadd bfloat %[[BF16_LOAD]], -1.000000e+00
+// OGCG:   store bfloat %[[BF16_DEC]], ptr %[[BF16_F]], align 2
+// OGCG:   ret bfloat %[[BF16_LOAD]]
+
 void test_logical_not() {
   int a = 5;
   a = !a;
@@ -481,28 +1588,28 @@ void test_logical_not() {
   b = !d;
 }
 
-// CHECK: cir.func{{.*}} @_Z16test_logical_notv()
-// CHECK:   %[[A:.*]] = cir.load{{.*}} %[[A_ADDR:.*]] : !cir.ptr<!s32i>, !s32i
-// CHECK:   %[[A_BOOL:.*]] = cir.cast int_to_bool %[[A]] : !s32i -> !cir.bool
-// CHECK:   %[[A_NOT:.*]] = cir.not %[[A_BOOL]] : !cir.bool
-// CHECK:   %[[A_CAST:.*]] = cir.cast bool_to_int %[[A_NOT]] : !cir.bool -> !s32i
-// CHECK:   cir.store{{.*}} %[[A_CAST]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i>
-// CHECK:   %[[B:.*]] = cir.load{{.*}} %[[B_ADDR:.*]] : !cir.ptr<!cir.bool>, !cir.bool
-// CHECK:   %[[B_NOT:.*]] = cir.not %[[B]] : !cir.bool
-// CHECK:   cir.store{{.*}} %[[B_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
-// CHECK:   %[[C:.*]] = cir.load{{.*}} %[[C_ADDR:.*]] : !cir.ptr<!cir.float>, !cir.float
-// CHECK:   %[[C_BOOL:.*]] = cir.cast float_to_bool %[[C]] : !cir.float -> !cir.bool
-// CHECK:   %[[C_NOT:.*]] = cir.not %[[C_BOOL]] : !cir.bool
-// CHECK:   %[[C_CAST:.*]] = cir.cast bool_to_float %[[C_NOT]] : !cir.bool -> !cir.float
-// CHECK:   cir.store{{.*}} %[[C_CAST]], %[[C_ADDR]] : !cir.float, !cir.ptr<!cir.float>
-// CHECK:   %[[P:.*]] = cir.load{{.*}} %[[P_ADDR:.*]] : !cir.ptr<!cir.ptr<!s32i>>, !cir.ptr<!s32i>
-// CHECK:   %[[P_BOOL:.*]] = cir.cast ptr_to_bool %[[P]] : !cir.ptr<!s32i> -> !cir.bool
-// CHECK:   %[[P_NOT:.*]] = cir.not %[[P_BOOL]] : !cir.bool
-// CHECK:   cir.store{{.*}} %[[P_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
-// CHECK:   %[[D:.*]] = cir.load{{.*}} %[[D_ADDR:.*]] : !cir.ptr<!cir.double>, !cir.double
-// CHECK:   %[[D_BOOL:.*]] = cir.cast float_to_bool %[[D]] : !cir.double -> !cir.bool
-// CHECK:   %[[D_NOT:.*]] = cir.not %[[D_BOOL]] : !cir.bool
-// CHECK:   cir.store{{.*}} %[[D_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
+// CIR: cir.func{{.*}} @_Z16test_logical_notv()
+// CIR:    %[[A:.*]] = cir.load{{.*}} %[[A_ADDR:.*]] : !cir.ptr<!s32i>, !s32i
+// CIR:    %[[A_BOOL:.*]] = cir.cast int_to_bool %[[A]] : !s32i -> !cir.bool
+// CIR:    %[[A_NOT:.*]] = cir.not %[[A_BOOL]] : !cir.bool
+// CIR:    %[[A_CAST:.*]] = cir.cast bool_to_int %[[A_NOT]] : !cir.bool -> !s32i
+// CIR:    cir.store{{.*}} %[[A_CAST]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i>
+// CIR:    %[[B:.*]] = cir.load{{.*}} %[[B_ADDR:.*]] : !cir.ptr<!cir.bool>, !cir.bool
+// CIR:    %[[B_NOT:.*]] = cir.not %[[B]] : !cir.bool
+// CIR:    cir.store{{.*}} %[[B_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
+// CIR:    %[[C:.*]] = cir.load{{.*}} %[[C_ADDR:.*]] : !cir.ptr<!cir.float>, !cir.float
+// CIR:    %[[C_BOOL:.*]] = cir.cast float_to_bool %[[C]] : !cir.float -> !cir.bool
+// CIR:    %[[C_NOT:.*]] = cir.not %[[C_BOOL]] : !cir.bool
+// CIR:    %[[C_CAST:.*]] = cir.cast bool_to_float %[[C_NOT]] : !cir.bool -> !cir.float
+// CIR:    cir.store{{.*}} %[[C_CAST]], %[[C_ADDR]] : !cir.float, !cir.ptr<!cir.float>
+// CIR:    %[[P:.*]] = cir.load{{.*}} %[[P_ADDR:.*]] : !cir.ptr<!cir.ptr<!s32i>>, !cir.ptr<!s32i>
+// CIR:    %[[P_BOOL:.*]] = cir.cast ptr_to_bool %[[P]] : !cir.ptr<!s32i> -> !cir.bool
+// CIR:    %[[P_NOT:.*]] = cir.not %[[P_BOOL]] : !cir.bool
+// CIR:    cir.store{{.*}} %[[P_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
+// CIR:    %[[D:.*]] = cir.load{{.*}} %[[D_ADDR:.*]] : !cir.ptr<!cir.double>, !cir.double
+// CIR:    %[[D_BOOL:.*]] = cir.cast float_to_bool %[[D]] : !cir.double -> !cir.bool
+// CIR:    %[[D_NOT:.*]] = cir.not %[[D_BOOL]] : !cir.bool
+// CIR:    cir.store{{.*}} %[[D_NOT]], %[[B_ADDR]] : !cir.bool, !cir.ptr<!cir.bool>
 
 // LLVM: define{{.*}} void @_Z16test_logical_notv()
 // LLVM:   %[[A:.*]] = load i32, ptr %[[A_ADDR:.*]], align 4
@@ -563,13 +1670,13 @@ void f16NestedUPlus() {
   _Float16 b = +(+a);
 }
 
-// CHECK: cir.func{{.*}} @_Z14f16NestedUPlusv()
-// CHECK:  %[[A_ADDR:.*]] = cir.alloca "a" {{.*}} : !cir.ptr<!cir.f16>
-// CHECK:  %[[B_ADDR:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.f16>
-// CHECK:  %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.f16>, !cir.f16
-// CHECK:  %[[A_F32:.*]] = cir.cast floating %[[TMP_A]] : !cir.f16 -> !cir.float
-// CHECK:  %[[RESULT:.*]] = cir.cast floating %[[A_F32]] : !cir.float -> !cir.f16
-// CHECK:  cir.store{{.*}} %[[RESULT]], %[[B_ADDR]] : !cir.f16, !cir.ptr<!cir.f16>
+// CIR: cir.func{{.*}} @_Z14f16NestedUPlusv()
+// CIR:   %[[A_ADDR:.*]] = cir.alloca "a" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:   %[[B_ADDR:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:   %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:   %[[A_F32:.*]] = cir.cast floating %[[TMP_A]] : !cir.f16 -> !cir.float
+// CIR:   %[[RESULT:.*]] = cir.cast floating %[[A_F32]] : !cir.float -> !cir.f16
+// CIR:   cir.store{{.*}} %[[RESULT]], %[[B_ADDR]] : !cir.f16, !cir.ptr<!cir.f16>
 
 // LLVM: define{{.*}} void @_Z14f16NestedUPlusv()
 // LLVM:  %[[A_ADDR:.*]] = alloca half, align 2
@@ -592,15 +1699,15 @@ void f16NestedUMinus() {
   _Float16 b = -(-a);
 }
 
-// CHECK: cir.func{{.*}} @_Z15f16NestedUMinusv()
-// CHECK:  %[[A_ADDR:.*]] = cir.alloca "a" {{.*}} : !cir.ptr<!cir.f16>
-// CHECK:  %[[B_ADDR:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.f16>
-// CHECK:  %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.f16>, !cir.f16
-// CHECK:  %[[A_F32:.*]] = cir.cast floating %[[TMP_A]] : !cir.f16 -> !cir.float
-// CHECK:  %[[A_MINUS:.*]] = cir.fneg %[[A_F32]] : !cir.float
-// CHECK:  %[[RESULT_F32:.*]] = cir.fneg %[[A_MINUS]] : !cir.float
-// CHECK:  %[[RESULT:.*]] = cir.cast floating %[[RESULT_F32]] : !cir.float -> !cir.f16
-// CHECK:  cir.store{{.*}} %[[RESULT]], %[[B_ADDR]] : !cir.f16, !cir.ptr<!cir.f16>
+// CIR: cir.func{{.*}} @_Z15f16NestedUMinusv()
+// CIR:   %[[A_ADDR:.*]] = cir.alloca "a" {{.*}} : !cir.ptr<!cir.f16>
+// CIR:   %[[B_ADDR:.*]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.f16>
+// CIR:   %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!cir.f16>, !cir.f16
+// CIR:   %[[A_F32:.*]] = cir.cast floating %[[TMP_A]] : !cir.f16 -> !cir.float
+// CIR:   %[[A_MINUS:.*]] = cir.fneg %[[A_F32]] : !cir.float
+// CIR:   %[[RESULT_F32:.*]] = cir.fneg %[[A_MINUS]] : !cir.float
+// CIR:   %[[RESULT:.*]] = cir.cast floating %[[RESULT_F32]] : !cir.float -> !cir.f16
+// CIR:   cir.store{{.*}} %[[RESULT]], %[[B_ADDR]] : !cir.f16, !cir.ptr<!cir.f16>
 
 // LLVM: define{{.*}} void @_Z15f16NestedUMinusv()
 // LLVM:  %[[A_ADDR:.*]] = alloca half, align 2
@@ -621,3 +1728,650 @@ void f16NestedUMinus() {
 // OGCG:  %[[RESULT_F32:.*]] = fneg float %[[A_MINUS]]
 // OGCG:  %[[RESULT:.*]] = fptrunc float %[[RESULT_F32]] to half
 // OGCG:  store half %[[RESULT]], ptr %[[B_ADDR]], align 2
+
+// Unary inc/dec on vector types, covering the CIRGenExprScalar.cpp vector
+// branch:
+//   * integer vectors go through emitIntIncOrDec -> cir.inc/cir.dec
+//   * float vectors go through emitFloatIncOrDec -> cir.fadd
+
+typedef int vi4 __attribute__((vector_size(16)));
+typedef unsigned uvi4 __attribute__((vector_size(16)));
+typedef short vsh8 __attribute__((vector_size(16)));
+typedef float vf4 __attribute__((vector_size(16)));
+typedef double vd2 __attribute__((vector_size(16)));
+
+vi4 vecIntPreInc(vi4 a) {
+  return ++a;
+}
+// CIR-LABEL: @_Z12vecIntPreIncDv4_i
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<4 x !s32i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:  cir.store %[[INCREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !s32i>
+
+// LLVM-LABEL: @_Z12vecIntPreIncDv4_i
+// LLVM: %[[A:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[A]]
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z12vecIntPreIncDv4_i
+// OGCG: %[[A:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// OGCG: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// OGCG: ret <4 x i32> %[[INCREMENTED]]
+
+vi4 vecIntPreDec(vi4 a) {
+  return --a;
+}
+// CIR-LABEL: @_Z12vecIntPreDecDv4_i
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<4 x !s32i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:  cir.store %[[DECREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !s32i>
+
+// LLVM-LABEL: @_Z12vecIntPreDecDv4_i
+// LLVM: %[[A:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[A]]
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z12vecIntPreDecDv4_i
+// OGCG: %[[A:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// OGCG: %[[DECREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 -1)
+// OGCG: ret <4 x i32> %[[DECREMENTED]]
+
+vi4 vecIntPostInc(vi4 a) {
+  return a++;
+}
+// CIR-LABEL: @_Z13vecIntPostIncDv4_i
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<4 x !s32i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !s32i>
+
+// LLVM-LABEL: @_Z13vecIntPostIncDv4_i
+// LLVM: %[[A:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[A]]
+// LLVM: store <4 x i32> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z13vecIntPostIncDv4_i
+// OGCG: %[[A:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// OGCG: add <4 x i32> %[[INPUT]], splat (i32 1)
+// OGCG: ret <4 x i32> %[[INPUT]]
+
+vi4 vecIntPostDec(vi4 a) {
+  return a--;
+}
+// CIR-LABEL: @_Z13vecIntPostDecDv4_i
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<4 x !s32i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !s32i>
+
+// LLVM-LABEL: @_Z13vecIntPostDecDv4_i
+// LLVM: %[[A:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[A]]
+// LLVM: store <4 x i32> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z13vecIntPostDecDv4_i
+// OGCG: %[[A:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[A]], align 16
+// OGCG: add <4 x i32> %[[INPUT]], splat (i32 -1)
+// OGCG: ret <4 x i32> %[[INPUT]]
+
+uvi4 vecUIntPreInc(uvi4 b) {
+  return ++b;
+}
+// CIR-LABEL: @_Z13vecUIntPreIncDv4_j
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<4 x !u32i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[B]]
+// CIR:  cir.store %[[INCREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !u32i>
+
+// LLVM-LABEL: @_Z13vecUIntPreIncDv4_j
+// LLVM: %[[B:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[B]]
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z13vecUIntPreIncDv4_j
+// OGCG: %[[B:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// OGCG: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// OGCG: ret <4 x i32> %[[INCREMENTED]]
+
+uvi4 vecUIntPreDec(uvi4 b) {
+  return --b;
+}
+// CIR-LABEL: @_Z13vecUIntPreDecDv4_j
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<4 x !u32i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[B]]
+// CIR:  cir.store %[[DECREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !u32i>
+
+// LLVM-LABEL: @_Z13vecUIntPreDecDv4_j
+// LLVM: %[[B:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[B]]
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z13vecUIntPreDecDv4_j
+// OGCG: %[[B:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// OGCG: %[[DECREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 -1)
+// OGCG: ret <4 x i32> %[[DECREMENTED]]
+
+uvi4 vecUIntPostInc(uvi4 b) {
+  return b++;
+}
+// CIR-LABEL: @_Z14vecUIntPostIncDv4_j
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<4 x !u32i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[B]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !u32i>
+
+// LLVM-LABEL: @_Z14vecUIntPostIncDv4_j
+// LLVM: %[[B:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[INCREMENTED]], ptr %[[B]]
+// LLVM: store <4 x i32> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecUIntPostIncDv4_j
+// OGCG: %[[B:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// OGCG: add <4 x i32> %[[INPUT]], splat (i32 1)
+// OGCG: ret <4 x i32> %[[INPUT]]
+
+uvi4 vecUIntPostDec(uvi4 b) {
+  return b--;
+}
+// CIR-LABEL: @_Z14vecUIntPostDecDv4_j
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<4 x !u32i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[B]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<4 x !u32i>
+
+// LLVM-LABEL: @_Z14vecUIntPostDecDv4_j
+// LLVM: %[[B:.+]] = alloca <4 x i32>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x i32>, align 16
+// LLVM: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <4 x i32> %[[INPUT]], splat (i32 1)
+// LLVM: store <4 x i32> %[[DECREMENTED]], ptr %[[B]]
+// LLVM: store <4 x i32> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x i32>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x i32> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecUIntPostDecDv4_j
+// OGCG: %[[B:.+]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x i32>, ptr %[[B]], align 16
+// OGCG: add <4 x i32> %[[INPUT]], splat (i32 -1)
+// OGCG: ret <4 x i32> %[[INPUT]]
+
+vsh8 vecShortPreInc(vsh8 c) {
+  return ++c;
+}
+// CIR-LABEL: @_Z14vecShortPreIncDv8_s
+// CIR:  %[[C:.+]] = cir.alloca "c" {{.*}} init : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[C]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<8 x !s16i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[C]]
+// CIR:  cir.store %[[INCREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<8 x !s16i>>, !cir.vector<8 x !s16i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<8 x !s16i>
+
+// LLVM-LABEL: @_Z14vecShortPreIncDv8_s
+// LLVM: %[[C:.+]] = alloca <8 x i16>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <8 x i16>, align 16
+// LLVM: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// LLVM: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <8 x i16> %[[INPUT]], splat (i16 1)
+// LLVM: store <8 x i16> %[[INCREMENTED]], ptr %[[C]]
+// LLVM: store <8 x i16> %[[INCREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <8 x i16>, ptr %[[RVPTR]], align 16
+// LLVM: ret <8 x i16> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecShortPreIncDv8_s
+// OGCG: %[[C:.+]] = alloca <8 x i16>, align 16
+// OGCG: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// OGCG: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// OGCG: %[[INCREMENTED:.+]] = add <8 x i16> %[[INPUT]], splat (i16 1)
+// OGCG: ret <8 x i16> %[[INCREMENTED]]
+
+vsh8 vecShortPreDec(vsh8 c) {
+  return --c;
+}
+// CIR-LABEL: @_Z14vecShortPreDecDv8_s
+// CIR:  %[[C:.+]] = cir.alloca "c" {{.*}} init : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[C]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<8 x !s16i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[C]]
+// CIR:  cir.store %[[DECREMENTED]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<8 x !s16i>>, !cir.vector<8 x !s16i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<8 x !s16i>
+
+// LLVM-LABEL: @_Z14vecShortPreDecDv8_s
+// LLVM: %[[C:.+]] = alloca <8 x i16>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <8 x i16>, align 16
+// LLVM: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// LLVM: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <8 x i16> %[[INPUT]], splat (i16 1)
+// LLVM: store <8 x i16> %[[DECREMENTED]], ptr %[[C]]
+// LLVM: store <8 x i16> %[[DECREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <8 x i16>, ptr %[[RVPTR]], align 16
+// LLVM: ret <8 x i16> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecShortPreDecDv8_s
+// OGCG: %[[C:.+]] = alloca <8 x i16>, align 16
+// OGCG: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// OGCG: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// OGCG: %[[DECREMENTED:.+]] = add <8 x i16> %[[INPUT]], splat (i16 -1)
+// OGCG: ret <8 x i16> %[[DECREMENTED]]
+
+vsh8 vecShortPostInc(vsh8 c) {
+  return c++;
+}
+// CIR-LABEL: @_Z15vecShortPostIncDv8_s
+// CIR:  %[[C:.+]] = cir.alloca "c" {{.*}} init : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[C]]
+// CIR:  %[[INCREMENTED:.+]] = cir.inc %[[INPUT]] : !cir.vector<8 x !s16i>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[C]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<8 x !s16i>>, !cir.vector<8 x !s16i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<8 x !s16i>
+
+// LLVM-LABEL: @_Z15vecShortPostIncDv8_s
+// LLVM: %[[C:.+]] = alloca <8 x i16>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <8 x i16>, align 16
+// LLVM: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// LLVM: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// LLVM: %[[INCREMENTED:.+]] = add <8 x i16> %[[INPUT]], splat (i16 1)
+// LLVM: store <8 x i16> %[[INCREMENTED]], ptr %[[C]]
+// LLVM: store <8 x i16> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <8 x i16>, ptr %[[RVPTR]], align 16
+// LLVM: ret <8 x i16> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecShortPostIncDv8_s
+// OGCG: %[[C:.+]] = alloca <8 x i16>, align 16
+// OGCG: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// OGCG: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// OGCG: add <8 x i16> %[[INPUT]], splat (i16 1)
+// OGCG: ret <8 x i16> %[[INPUT]]
+
+vsh8 vecShortPostDec(vsh8 c) {
+  return c--;
+}
+// CIR-LABEL: @_Z15vecShortPostDecDv8_s
+// CIR:  %[[C:.+]] = cir.alloca "c" {{.*}} init : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[RV:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<8 x !s16i>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[C]]
+// CIR:  %[[DECREMENTED:.+]] = cir.dec %[[INPUT]] : !cir.vector<8 x !s16i>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[C]]
+// CIR:  cir.store %[[INPUT]], %[[RV]]
+// CIR:  %[[RV_LOAD:.+]] = cir.load %[[RV]] : !cir.ptr<!cir.vector<8 x !s16i>>, !cir.vector<8 x !s16i>
+// CIR:  cir.return %[[RV_LOAD]] : !cir.vector<8 x !s16i>
+
+// LLVM-LABEL: @_Z15vecShortPostDecDv8_s
+// LLVM: %[[C:.+]] = alloca <8 x i16>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <8 x i16>, align 16
+// LLVM: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// LLVM: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// LLVM: %[[DECREMENTED:.+]] = sub <8 x i16> %[[INPUT]], splat (i16 1)
+// LLVM: store <8 x i16> %[[DECREMENTED]], ptr %[[C]]
+// LLVM: store <8 x i16> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <8 x i16>, ptr %[[RVPTR]], align 16
+// LLVM: ret <8 x i16> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecShortPostDecDv8_s
+// OGCG: %[[C:.+]] = alloca <8 x i16>, align 16
+// OGCG: store <8 x i16> %{{.+}}, ptr %[[C]], align 16
+// OGCG: %[[INPUT:.+]] = load <8 x i16>, ptr %[[C]], align 16
+// OGCG: add <8 x i16> %[[INPUT]], splat (i16 -1)
+// OGCG: ret <8 x i16> %[[INPUT]]
+
+vf4 vecFloatPreInc(vf4 a) {
+  return ++a;
+}
+// CIR-LABEL: @_Z14vecFloatPreIncDv4_f
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:  %[[ONEVEC:.*]] = cir.vec.splat %[[ONE]] : !cir.float, !cir.vector<4 x !cir.float>
+// CIR:  %[[INCREMENTED:.+]] = cir.fadd %[[INPUT]], %[[ONEVEC]] : !cir.vector<4 x !cir.float>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:  cir.store %[[INCREMENTED]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<4 x !cir.float>>, !cir.vector<4 x !cir.float>
+// CIR:  cir.return %[[RV]] : !cir.vector<4 x !cir.float>
+
+// LLVM-LABEL: @_Z14vecFloatPreIncDv4_f
+// LLVM: %[[A:.+]] = alloca <4 x float>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x float>, align 16
+// LLVM: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// LLVM: %[[INCREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float 1.000000e+00)
+// LLVM: store <4 x float> %[[INCREMENTED]], ptr %[[A]]
+// LLVM: store <4 x float> %[[INCREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x float>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x float> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecFloatPreIncDv4_f
+// OGCG: %[[A:.+]] = alloca <4 x float>, align 16
+// OGCG: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// OGCG: %[[INCREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float 1.000000e+00)
+// OGCG: ret <4 x float> %[[INCREMENTED]]
+
+vf4 vecFloatPreDec(vf4 a) {
+  return --a;
+}
+// CIR-LABEL: @_Z14vecFloatPreDecDv4_f
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:  %[[NEGONEVEC:.*]] = cir.vec.splat %[[NEGONE]] : !cir.float, !cir.vector<4 x !cir.float>
+// CIR:  %[[DECREMENTED:.+]] = cir.fadd %[[INPUT]], %[[NEGONEVEC]] : !cir.vector<4 x !cir.float>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:  cir.store %[[DECREMENTED]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<4 x !cir.float>>, !cir.vector<4 x !cir.float>
+// CIR:  cir.return %[[RV]] : !cir.vector<4 x !cir.float>
+
+// LLVM-LABEL: @_Z14vecFloatPreDecDv4_f
+// LLVM: %[[A:.+]] = alloca <4 x float>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x float>, align 16
+// LLVM: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// LLVM: %[[DECREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float -1.000000e+00)
+// LLVM: store <4 x float> %[[DECREMENTED]], ptr %[[A]]
+// LLVM: store <4 x float> %[[DECREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x float>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x float> %[[RV]]
+
+// OGCG-LABEL: @_Z14vecFloatPreDecDv4_f
+// OGCG: %[[A:.+]] = alloca <4 x float>, align 16
+// OGCG: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// OGCG: %[[DECREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float -1.000000e+00)
+// OGCG: ret <4 x float> %[[DECREMENTED]]
+
+vf4 vecFloatPostInc(vf4 a) {
+  return a++;
+}
+// CIR-LABEL: @_Z15vecFloatPostIncDv4_f
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
+// CIR:  %[[ONEVEC:.*]] = cir.vec.splat %[[ONE]] : !cir.float, !cir.vector<4 x !cir.float>
+// CIR:  %[[INCREMENTED:.+]] = cir.fadd %[[INPUT]], %[[ONEVEC]] : !cir.vector<4 x !cir.float>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[A]]
+// CIR:  cir.store %[[INPUT]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<4 x !cir.float>>, !cir.vector<4 x !cir.float>
+// CIR:  cir.return %[[RV]] : !cir.vector<4 x !cir.float>
+
+// LLVM-LABEL: @_Z15vecFloatPostIncDv4_f
+// LLVM: %[[A:.+]] = alloca <4 x float>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x float>, align 16
+// LLVM: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// LLVM: %[[INCREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float 1.000000e+00)
+// LLVM: store <4 x float> %[[INCREMENTED]], ptr %[[A]]
+// LLVM: store <4 x float> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x float>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x float> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecFloatPostIncDv4_f
+// OGCG: %[[A:.+]] = alloca <4 x float>, align 16
+// OGCG: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// OGCG: fadd <4 x float> %[[INPUT]], splat (float 1.000000e+00)
+// OGCG: ret <4 x float> %[[INPUT]]
+
+vf4 vecFloatPostDec(vf4 a) {
+  return a--;
+}
+// CIR-LABEL: @_Z15vecFloatPostDecDv4_f
+// CIR:  %[[A:.+]] = cir.alloca "a" {{.*}} init : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<4 x !cir.float>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[A]]
+// CIR:  %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+// CIR:  %[[NEGONEVEC:.*]] = cir.vec.splat %[[NEGONE]] : !cir.float, !cir.vector<4 x !cir.float>
+// CIR:  %[[DECREMENTED:.+]] = cir.fadd %[[INPUT]], %[[NEGONEVEC]] : !cir.vector<4 x !cir.float>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[A]]
+// CIR:  cir.store %[[INPUT]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<4 x !cir.float>>, !cir.vector<4 x !cir.float>
+// CIR:  cir.return %[[RV]] : !cir.vector<4 x !cir.float>
+
+// LLVM-LABEL: @_Z15vecFloatPostDecDv4_f
+// LLVM: %[[A:.+]] = alloca <4 x float>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <4 x float>, align 16
+// LLVM: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// LLVM: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// LLVM: %[[DECREMENTED:.+]] = fadd <4 x float> %[[INPUT]], splat (float -1.000000e+00)
+// LLVM: store <4 x float> %[[DECREMENTED]], ptr %[[A]]
+// LLVM: store <4 x float> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <4 x float>, ptr %[[RVPTR]], align 16
+// LLVM: ret <4 x float> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecFloatPostDecDv4_f
+// OGCG: %[[A:.+]] = alloca <4 x float>, align 16
+// OGCG: store <4 x float> %{{.+}}, ptr %[[A]], align 16
+// OGCG: %[[INPUT:.+]] = load <4 x float>, ptr %[[A]], align 16
+// OGCG: fadd <4 x float> %[[INPUT]], splat (float -1.000000e+00)
+// OGCG: ret <4 x float> %[[INPUT]]
+
+vd2 vecDoublePreInc(vd2 b) {
+  return ++b;
+}
+// CIR-LABEL: @_Z15vecDoublePreIncDv2_d
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.double
+// CIR:  %[[ONEVEC:.*]] = cir.vec.splat %[[ONE]] : !cir.double, !cir.vector<2 x !cir.double>
+// CIR:  %[[INCREMENTED:.+]] = cir.fadd %[[INPUT]], %[[ONEVEC]] : !cir.vector<2 x !cir.double>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[B]]
+// CIR:  cir.store %[[INCREMENTED]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<2 x !cir.double>>, !cir.vector<2 x !cir.double>
+// CIR:  cir.return %[[RV]] : !cir.vector<2 x !cir.double>
+
+// LLVM-LABEL: @_Z15vecDoublePreIncDv2_d
+// LLVM: %[[B:.+]] = alloca <2 x double>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <2 x double>, align 16
+// LLVM: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// LLVM: %[[INCREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double 1.000000e+00)
+// LLVM: store <2 x double> %[[INCREMENTED]], ptr %[[B]]
+// LLVM: store <2 x double> %[[INCREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <2 x double>, ptr %[[RVPTR]], align 16
+// LLVM: ret <2 x double> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecDoublePreIncDv2_d
+// OGCG: %[[B:.+]] = alloca <2 x double>, align 16
+// OGCG: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// OGCG: %[[INCREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double 1.000000e+00)
+// OGCG: ret <2 x double> %[[INCREMENTED]]
+
+vd2 vecDoublePreDec(vd2 b) {
+  return --b;
+}
+// CIR-LABEL: @_Z15vecDoublePreDecDv2_d
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.double
+// CIR:  %[[NEGONEVEC:.*]] = cir.vec.splat %[[NEGONE]] : !cir.double, !cir.vector<2 x !cir.double>
+// CIR:  %[[DECREMENTED:.+]] = cir.fadd %[[INPUT]], %[[NEGONEVEC]] : !cir.vector<2 x !cir.double>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[B]]
+// CIR:  cir.store %[[DECREMENTED]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<2 x !cir.double>>, !cir.vector<2 x !cir.double>
+// CIR:  cir.return %[[RV]] : !cir.vector<2 x !cir.double>
+
+// LLVM-LABEL: @_Z15vecDoublePreDecDv2_d
+// LLVM: %[[B:.+]] = alloca <2 x double>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <2 x double>, align 16
+// LLVM: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// LLVM: %[[DECREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double -1.000000e+00)
+// LLVM: store <2 x double> %[[DECREMENTED]], ptr %[[B]]
+// LLVM: store <2 x double> %[[DECREMENTED]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <2 x double>, ptr %[[RVPTR]], align 16
+// LLVM: ret <2 x double> %[[RV]]
+
+// OGCG-LABEL: @_Z15vecDoublePreDecDv2_d
+// OGCG: %[[B:.+]] = alloca <2 x double>, align 16
+// OGCG: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// OGCG: %[[DECREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double -1.000000e+00)
+// OGCG: ret <2 x double> %[[DECREMENTED]]
+
+vd2 vecDoublePostInc(vd2 b) {
+  return b++;
+}
+// CIR-LABEL: @_Z16vecDoublePostIncDv2_d
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.double
+// CIR:  %[[ONEVEC:.*]] = cir.vec.splat %[[ONE]] : !cir.double, !cir.vector<2 x !cir.double>
+// CIR:  %[[INCREMENTED:.+]] = cir.fadd %[[INPUT]], %[[ONEVEC]] : !cir.vector<2 x !cir.double>
+// CIR:  cir.store{{.*}} %[[INCREMENTED]], %[[B]]
+// CIR:  cir.store %[[INPUT]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<2 x !cir.double>>, !cir.vector<2 x !cir.double>
+// CIR:  cir.return %[[RV]] : !cir.vector<2 x !cir.double>
+
+// LLVM-LABEL: @_Z16vecDoublePostIncDv2_d
+// LLVM: %[[B:.+]] = alloca <2 x double>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <2 x double>, align 16
+// LLVM: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// LLVM: %[[INCREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double 1.000000e+00)
+// LLVM: store <2 x double> %[[INCREMENTED]], ptr %[[B]]
+// LLVM: store <2 x double> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <2 x double>, ptr %[[RVPTR]], align 16
+// LLVM: ret <2 x double> %[[RV]]
+
+// OGCG-LABEL: @_Z16vecDoublePostIncDv2_d
+// OGCG: %[[B:.+]] = alloca <2 x double>, align 16
+// OGCG: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// OGCG: fadd <2 x double> %[[INPUT]], splat (double 1.000000e+00)
+// OGCG: ret <2 x double> %[[INPUT]]
+
+vd2 vecDoublePostDec(vd2 b) {
+  return b--;
+}
+// CIR-LABEL: @_Z16vecDoublePostDecDv2_d
+// CIR:  %[[B:.+]] = cir.alloca "b" {{.*}} init : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[RVPTR:.+]] = cir.alloca "__retval" {{.*}} : !cir.ptr<!cir.vector<2 x !cir.double>>
+// CIR:  %[[INPUT:.+]] = cir.load{{.*}} %[[B]]
+// CIR:  %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.double
+// CIR:  %[[NEGONEVEC:.*]] = cir.vec.splat %[[NEGONE]] : !cir.double, !cir.vector<2 x !cir.double>
+// CIR:  %[[DECREMENTED:.+]] = cir.fadd %[[INPUT]], %[[NEGONEVEC]] : !cir.vector<2 x !cir.double>
+// CIR:  cir.store{{.*}} %[[DECREMENTED]], %[[B]]
+// CIR:  cir.store %[[INPUT]], %[[RVPTR]]
+// CIR:  %[[RV:.+]] = cir.load %[[RVPTR]] : !cir.ptr<!cir.vector<2 x !cir.double>>, !cir.vector<2 x !cir.double>
+// CIR:  cir.return %[[RV]] : !cir.vector<2 x !cir.double>
+
+// LLVM-LABEL: @_Z16vecDoublePostDecDv2_d
+// LLVM: %[[B:.+]] = alloca <2 x double>, align 16
+// LLVM: %[[RVPTR:.+]] = alloca <2 x double>, align 16
+// LLVM: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// LLVM: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// LLVM: %[[DECREMENTED:.+]] = fadd <2 x double> %[[INPUT]], splat (double -1.000000e+00)
+// LLVM: store <2 x double> %[[DECREMENTED]], ptr %[[B]]
+// LLVM: store <2 x double> %[[INPUT]], ptr %[[RVPTR]]
+// LLVM: %[[RV:.+]] = load <2 x double>, ptr %[[RVPTR]], align 16
+// LLVM: ret <2 x double> %[[RV]]
+
+// OGCG-LABEL: @_Z16vecDoublePostDecDv2_d
+// OGCG: %[[B:.+]] = alloca <2 x double>, align 16
+// OGCG: store <2 x double> %{{.+}}, ptr %[[B]], align 16
+// OGCG: %[[INPUT:.+]] = load <2 x double>, ptr %[[B]], align 16
+// OGCG: fadd <2 x double> %[[INPUT]], splat (double -1.000000e+00)
+// OGCG: ret <2 x double> %[[INPUT]]

diff  --git a/clang/test/CIR/CodeGenOpenACC/atomic-update.cpp b/clang/test/CIR/CodeGenOpenACC/atomic-update.cpp
index da511723489e9..353db84000c02 100644
--- a/clang/test/CIR/CodeGenOpenACC/atomic-update.cpp
+++ b/clang/test/CIR/CodeGenOpenACC/atomic-update.cpp
@@ -53,8 +53,8 @@ void use(int x, unsigned int y, float f, HasOps ops) {
   // CHECK-NEXT: cir.store %[[RECIPE_ARG]], %[[TEMP_ALLOCA]] : !cir.float, !cir.ptr<!cir.float>
   //
   // CHECK-NEXT: %[[TEMP_LOAD:.*]] = cir.load{{.*}} %[[TEMP_ALLOCA]] : !cir.ptr<!cir.float>, !cir.float
-  // CHECK-NEXT: %[[ONE:.*]] = cir.const #cir.fp<1.000000e+00> : !cir.float
-  // CHECK-NEXT: %[[INC:.*]] = cir.fsub %[[TEMP_LOAD]], %[[ONE]] : !cir.float
+  // CHECK-NEXT: %[[NEGONE:.*]] = cir.const #cir.fp<-1.000000e+00> : !cir.float
+  // CHECK-NEXT: %[[INC:.*]] = cir.fadd %[[TEMP_LOAD]], %[[NEGONE]] : !cir.float
   // CHECK-NEXT: cir.store {{.*}}%[[INC]], %[[TEMP_ALLOCA]] : !cir.float, !cir.ptr<!cir.float>
   // 
   // CHECK-NEXT: %[[TEMP_LOAD:.*]] = cir.load{{.*}} %[[TEMP_ALLOCA]] : !cir.ptr<!cir.float>, !cir.float


        


More information about the cfe-commits mailing list