From 30150a128c7b2488225331417153ccec75bac65c Mon Sep 17 00:00:00 2001
From: Nadav Rotem <nrotem@apple.com>
Date: Sun, 10 Nov 2013 04:13:31 +0000
Subject: SimplifyCFG has a heuristics for out-of-order processors that decides
 when it is worthwhile to merge branches. It tries to estimate if the operands
 of the instruction that we want to hoist are ready. This commit marks
 function arguments as 'ready' because they require no calculation. This
 boosts libquantum and a few other workloads from the testsuite.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194346 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Transforms/Utils/SimplifyCFG.cpp               |  2 +-
 test/Transforms/SimplifyCFG/common-dest-folding.ll | 28 ++++++++++++++++++++++
 2 files changed, 29 insertions(+), 1 deletion(-)
 create mode 100644 test/Transforms/SimplifyCFG/common-dest-folding.ll

diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp
index 1398697..d36d9dc 100644
--- a/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -2096,7 +2096,7 @@ bool llvm::FoldBranchToCommonDest(BranchInst *BI) {
       for (Instruction::op_iterator OI = BonusInst->op_begin(),
            OE = BonusInst->op_end(); OI != OE; ++OI) {
         Value *V = *OI;
-        if (!isa<Constant>(V))
+        if (!isa<Constant>(V) && !isa<Argument>(V))
           UsedValues.insert(V);
       }
 
diff --git a/test/Transforms/SimplifyCFG/common-dest-folding.ll b/test/Transforms/SimplifyCFG/common-dest-folding.ll
new file mode 100644
index 0000000..10fb7d4
--- /dev/null
+++ b/test/Transforms/SimplifyCFG/common-dest-folding.ll
@@ -0,0 +1,28 @@
+; RUN: opt < %s -simplifycfg -S | FileCheck %s
+
+;CHECK: @foo
+;CHECK: and i32 %c1, %k
+;CHECK: icmp eq i32
+;CHECK: and i32 %c2, %k
+;CHECK: icmp eq i32
+;CHECK: or i1
+;CHECK: ret
+define i32 @foo(i32 %k, i32 %c1, i32 %c2) {
+  %1 = and i32 %c1, %k
+  %2 = icmp eq i32 %1, 0
+  br i1 %2, label %8, label %3
+
+; <label>:3                                       ; preds = %0
+  %4 = and i32 %c2, %k
+  %5 = icmp eq i32 %4, 0
+  br i1 %5, label %8, label %6
+
+; <label>:6                                       ; preds = %3
+  %7 = tail call i32 (...)* @bar() nounwind
+  br label %8
+
+; <label>:8                                       ; preds = %3, %0, %6
+  ret i32 undef
+}
+
+declare i32 @bar(...)
-- 
cgit v1.1