From: Evan Cheng <evan.cheng@apple.com>
Date: Wed, 27 Apr 2011 19:32:43 +0000 (+0000)
Subject: If converter was being too cute. It look for root BBs (which don't have
X-Git-Url: http://plrg.eecs.uci.edu/git/?a=commitdiff_plain;h=309db7c947de7f0f9730dff5f95bc1cc5bd1db7d;p=oota-llvm.git

If converter was being too cute. It look for root BBs (which don't have
successors) and use inverse depth first search to traverse the BBs. However
that doesn't work when the CFG has infinite loops. Simply do a linear
traversal of all BBs work just fine.

rdar://9344645


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130324 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/CodeGen/IfConversion.cpp b/lib/CodeGen/IfConversion.cpp
index db53b0473a9..790200b8df5 100644
--- a/lib/CodeGen/IfConversion.cpp
+++ b/lib/CodeGen/IfConversion.cpp
@@ -27,7 +27,6 @@
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/raw_ostream.h"
-#include "llvm/ADT/DepthFirstIterator.h"
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/STLExtras.h"
@@ -146,10 +145,6 @@ namespace {
         : BBI(b), Kind(k), NeedSubsumption(s), NumDups(d), NumDups2(d2) {}
     };
 
-    /// Roots - Basic blocks that do not have successors. These are the starting
-    /// points of Graph traversal.
-    std::vector<MachineBasicBlock*> Roots;
-
     /// BBAnalysis - Results of if-conversion feasibility analysis indexed by
     /// basic block number.
     std::vector<BBInfo> BBAnalysis;
@@ -287,11 +282,6 @@ bool IfConverter::runOnMachineFunction(MachineFunction &MF) {
   MF.RenumberBlocks();
   BBAnalysis.resize(MF.getNumBlockIDs());
 
-  // Look for root nodes, i.e. blocks without successors.
-  for (MachineFunction::iterator I = MF.begin(), E = MF.end(); I != E; ++I)
-    if (I->succ_empty())
-      Roots.push_back(I);
-
   std::vector<IfcvtToken*> Tokens;
   MadeChange = false;
   unsigned NumIfCvts = NumSimple + NumSimpleFalse + NumTriangle +
@@ -406,7 +396,6 @@ bool IfConverter::runOnMachineFunction(MachineFunction &MF) {
   }
 
   Tokens.clear();
-  Roots.clear();
   BBAnalysis.clear();
 
   if (MadeChange && IfCvtBranchFold) {
@@ -924,13 +913,9 @@ IfConverter::BBInfo &IfConverter::AnalyzeBlock(MachineBasicBlock *BB,
 /// candidates.
 void IfConverter::AnalyzeBlocks(MachineFunction &MF,
                                 std::vector<IfcvtToken*> &Tokens) {
-  std::set<MachineBasicBlock*> Visited;
-  for (unsigned i = 0, e = Roots.size(); i != e; ++i) {
-    for (idf_ext_iterator<MachineBasicBlock*> I=idf_ext_begin(Roots[i],Visited),
-           E = idf_ext_end(Roots[i], Visited); I != E; ++I) {
-      MachineBasicBlock *BB = *I;
-      AnalyzeBlock(BB, Tokens);
-    }
+  for (MachineFunction::iterator I = MF.begin(), E = MF.end(); I != E; ++I) {
+    MachineBasicBlock *BB = I;
+    AnalyzeBlock(BB, Tokens);
   }
 
   // Sort to favor more complex ifcvt scheme.
diff --git a/test/CodeGen/ARM/2011-04-27-IfCvtBug.ll b/test/CodeGen/ARM/2011-04-27-IfCvtBug.ll
new file mode 100644
index 00000000000..0741049cffd
--- /dev/null
+++ b/test/CodeGen/ARM/2011-04-27-IfCvtBug.ll
@@ -0,0 +1,59 @@
+; RUN: llc < %s -mtriple=thumbv7-apple-ios
+
+; If converter was being too cute. It look for root BBs (which don't have
+; successors) and use inverse depth first search to traverse the BBs. However
+; that doesn't work when the CFG has infinite loops. Simply do a linear
+; traversal of all BBs work just fine.
+
+; rdar://9344645
+
+%struct.hc = type { i32, i32, i32, i32 }
+
+define i32 @t(i32 %type) optsize {
+entry:
+  br i1 undef, label %if.then, label %if.else
+
+if.then:
+  unreachable
+
+if.else:
+  br i1 undef, label %if.then15, label %if.else18
+
+if.then15:
+  unreachable
+
+if.else18:
+  switch i32 %type, label %if.else173 [
+    i32 3, label %if.then115
+    i32 1, label %if.then102
+  ]
+
+if.then102:
+  br i1 undef, label %cond.true10.i, label %t.exit
+
+cond.true10.i:
+  br label %t.exit
+
+t.exit:
+  unreachable
+
+if.then115:
+  br i1 undef, label %if.else163, label %if.else145
+
+if.else145:
+  %call150 = call fastcc %struct.hc* @foo(%struct.hc* undef, i32 34865152) optsize
+  br label %while.body172
+
+if.else163:
+  %call168 = call fastcc %struct.hc* @foo(%struct.hc* undef, i32 34078720) optsize
+  br label %while.body172
+
+while.body172:
+  br label %while.body172
+
+if.else173:
+  ret i32 -1
+}
+
+declare hidden fastcc %struct.hc* @foo(%struct.hc* nocapture, i32) nounwind optsize
+