[Clang] Pragma vectorize_width() implies vectorize(enable)

Specifying the vectorization width was supposed to implicitly enable
vectorization, except that it wasn't really doing this. It was only
setting the vectorize.width metadata, but not vectorize.enable.

This should fix PR27643.

Differential Revision: https://reviews.llvm.org/D66290

llvm-svn: 372082

1 files changed, 8 insertions, 0 deletions

diff --git a/clang/lib/CodeGen/CGLoopInfo.cpp b/clang/lib/CodeGen/CGLoopInfo.cpp
index 6822c6286fe..c21d4feee7a 100644
--- a/clang/lib/CodeGen/CGLoopInfo.cpp
+++ b/clang/lib/CodeGen/CGLoopInfo.cpp
@@ -270,6 +270,14 @@ LoopInfo::createLoopVectorizeMetadata(const LoopAttributes &Attrs,
 
   // Setting vectorize.width
   if (Attrs.VectorizeWidth > 0) {
+    // This implies vectorize.enable = true, but only add it when it is not
+    // already enabled.
+    if (Attrs.VectorizeEnable != LoopAttributes::Enable)
+      Args.push_back(
+          MDNode::get(Ctx, {MDString::get(Ctx, "llvm.loop.vectorize.enable"),
+                            ConstantAsMetadata::get(ConstantInt::get(
+                                llvm::Type::getInt1Ty(Ctx), 1))}));
+
     Metadata *Vals[] = {
         MDString::get(Ctx, "llvm.loop.vectorize.width"),
         ConstantAsMetadata::get(ConstantInt::get(llvm::Type::getInt32Ty(Ctx),