[llvm-branch-commits] [clang] 14947cd - [clang] Fix signedness in vector bitcast evaluation

Sven van Haastregt via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Mon Jan 25 04:05:56 PST 2021


Author: Sven van Haastregt
Date: 2021-01-25T12:01:42Z
New Revision: 14947cd04701d923a57a0161fd1967b81e00ff5e

URL: https://github.com/llvm/llvm-project/commit/14947cd04701d923a57a0161fd1967b81e00ff5e
DIFF: https://github.com/llvm/llvm-project/commit/14947cd04701d923a57a0161fd1967b81e00ff5e.diff

LOG: [clang] Fix signedness in vector bitcast evaluation

The included test case triggered a sign assertion on the result in
`Success()`.  This was caused by the APSInt created for a bitcast
having its signedness bit inverted.  The second APSInt constructor
argument is `isUnsigned`, so invert the result of
`isSignedIntegerType`.

Differential Revision: https://reviews.llvm.org/D95135

Added: 
    

Modified: 
    clang/lib/AST/ExprConstant.cpp
    clang/test/CodeGenOpenCL/vector_literals.cl

Removed: 
    


################################################################################
diff  --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index 56181bbe1166..c1973720e49a 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -10193,7 +10193,7 @@ bool VectorExprEvaluator::VisitCastExpr(const CastExpr *E) {
           Elt = SValInt.rotl(i*EltSize+EltSize).zextOrTrunc(EltSize);
         else
           Elt = SValInt.rotr(i*EltSize).zextOrTrunc(EltSize);
-        Elts.push_back(APValue(APSInt(Elt, EltTy->isSignedIntegerType())));
+        Elts.push_back(APValue(APSInt(Elt, !EltTy->isSignedIntegerType())));
       }
     } else {
       return Error(E);

diff  --git a/clang/test/CodeGenOpenCL/vector_literals.cl b/clang/test/CodeGenOpenCL/vector_literals.cl
index af571714368e..3115aef2b50a 100644
--- a/clang/test/CodeGenOpenCL/vector_literals.cl
+++ b/clang/test/CodeGenOpenCL/vector_literals.cl
@@ -63,3 +63,10 @@ void vector_literals_valid() {
   //CHECK: store <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>, <4 x float>* %V2
   float4 V2 = (float4)(1);
 }
+
+void vector_literals_with_cast() {
+  // CHECK-LABEL: vector_literals_with_cast
+  // CHECK: store <2 x i32> <i32 12, i32 34>, <2 x i32>*
+  // CHECK: extractelement <2 x i32> %{{[0-9]+}}, i64 0
+  unsigned int withCast = ((int2)((int2)(12, 34))).s0;
+}


        


More information about the llvm-branch-commits mailing list