[WideableCond] Fix a nasty bug in detection of "explicit guards"

The code was failing to actually check for the presence of the call to widenable_condition.  The whole point of specifying the widenable_condition intrinsic was allowing widening transforms.  A normal branch is not widenable.  A normal branch leading to a deopt is not widenable (in general).

I added a test case via LoopPredication, but GuardWidening has an analogous bug.  Those are the only two passes actually using this utility just yet. Noticed while working on LoopPredication for non-widenable branches; POC in D60111.

llvm-svn: 357493

1 files changed, 7 insertions, 2 deletions

diff --git a/llvm/lib/Analysis/GuardUtils.cpp b/llvm/lib/Analysis/GuardUtils.cpp
index c099ee8e41c..cad92f6e56b 100644
--- a/llvm/lib/Analysis/GuardUtils.cpp
+++ b/llvm/lib/Analysis/GuardUtils.cpp
@@ -39,6 +39,11 @@ bool llvm::parseWidenableBranch(const User *U, Value *&Condition,
                                 Value *&WidenableCondition,
                                 BasicBlock *&IfTrueBB, BasicBlock *&IfFalseBB) {
   using namespace llvm::PatternMatch;
-  return match(U, m_Br(m_And(m_Value(Condition), m_Value(WidenableCondition)),
-                       IfTrueBB, IfFalseBB));
+  if (!match(U, m_Br(m_And(m_Value(Condition), m_Value(WidenableCondition)),
+                     IfTrueBB, IfFalseBB)))
+    return false;
+  // TODO: At the moment, we only recognize the branch if the WC call in this
+  // specific position.  We should generalize!
+  return match(WidenableCondition,
+               m_Intrinsic<Intrinsic::experimental_widenable_condition>());
 }