[Clang][AArch64][SVE] Allow write to SVE vector elements using the subscript operator

[momchil-velikov]

The patch at https://reviews.llvm.org/D122732 introduced using the array subscript operator for SVE vectors, however it also causes an ICE when the subscripting expression is used as an lvalue.

This patches fixes the error. Lvalue subscripting expressions are emitted as LLVM IR insertelement.

[llvmbot]

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-clang-codegen

Author: Momchil Velikov (momchil-velikov)

Changes

The patch at https://reviews.llvm.org/D122732 introduced using the array subscript operator for SVE vectors, however it also causes an ICE when the subscripting expression is used as an lvalue.

This patches fixes the error. Lvalue subscripting expressions are emitted as LLVM IR insertelement.

---

Full diff: https://github.com/llvm/llvm-project/pull/91965.diff

3 Files Affected:

• (modified) clang/lib/CodeGen/CGExpr.cpp (+4-2)
• (modified) clang/lib/Sema/SemaExpr.cpp (+3-1)
• (modified) clang/test/CodeGen/aarch64-sve-vector-subscript-ops.c (+10)

diff --git a/clang/lib/CodeGen/CGExpr.cpp b/clang/lib/CodeGen/CGExpr.cpp
index d96c7bb1e5682..37b8b723937b7 100644
--- a/clang/lib/CodeGen/CGExpr.cpp
+++ b/clang/lib/CodeGen/CGExpr.cpp
@@ -4180,8 +4180,10 @@ LValue CodeGenFunction::EmitArraySubscriptExpr(const ArraySubscriptExpr *E,
 
   // If the base is a vector type, then we are forming a vector element lvalue
   // with this subscript.
-  if (E->getBase()->getType()->isVectorType() &&
-      !isa<ExtVectorElementExpr>(E->getBase())) {
+  if (QualType BaseTy = E->getBase()->getType();
+      (BaseTy->isVectorType() && !isa<ExtVectorElementExpr>(E->getBase())) ||
+      (BaseTy->isBuiltinType() &&
+       BaseTy->getAs<BuiltinType>()->isSveVLSBuiltinType())) {
     // Emit the vector as an lvalue to get its address.
     LValue LHS = EmitLValue(E->getBase());
     auto *Idx = EmitIdxAfterBase(/*Promote*/false);
diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
index bb4b116fd73ca..fd16be30bd848 100644
--- a/clang/lib/Sema/SemaExpr.cpp
+++ b/clang/lib/Sema/SemaExpr.cpp
@@ -5383,7 +5383,9 @@ Sema::CreateBuiltinArraySubscriptExpr(Expr *Base, SourceLocation LLoc,
   }
 
   // Perform default conversions.
-  if (!LHSExp->getType()->getAs<VectorType>()) {
+  if (!LHSExp->getType()->getAs<VectorType>() &&
+      !(LHSExp->getType()->isBuiltinType() &&
+        LHSExp->getType()->getAs<BuiltinType>()->isSveVLSBuiltinType())) {
     ExprResult Result = DefaultFunctionArrayLvalueConversion(LHSExp);
     if (Result.isInvalid())
       return ExprError();
diff --git a/clang/test/CodeGen/aarch64-sve-vector-subscript-ops.c b/clang/test/CodeGen/aarch64-sve-vector-subscript-ops.c
index fb60c6d100ce6..634423765c4c3 100644
--- a/clang/test/CodeGen/aarch64-sve-vector-subscript-ops.c
+++ b/clang/test/CodeGen/aarch64-sve-vector-subscript-ops.c
@@ -88,3 +88,13 @@ float subscript_float32(svfloat32_t a, size_t b) {
 double subscript_float64(svfloat64_t a, size_t b) {
   return a[b];
 }
+
+// CHECK-LABEL: @subscript_write_float32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[VECINS:%.*]] = insertelement <vscale x 4 x float> [[A:%.*]], float 1.000000e+00, i64 [[B:%.*]]
+// CHECK-NEXT:    ret <vscale x 4 x float> [[VECINS]]
+//
+svfloat32_t subscript_write_float32(svfloat32_t a, size_t b) {
+  a[b] = 1.0f;
+  return a;
+}

[DavidTruby]

LGTM

[momchil-velikov]

Rebased.

[efriedma-quic]

LGTM