From 403256763fc91a616a517a0ba0b128d28b8e166f Mon Sep 17 00:00:00 2001
From: Rafael Espindola <rafael.espindola@gmail.com>
Date: Mon, 11 Jul 2011 03:43:47 +0000
Subject: [PATCH] Don't duplicate the work done by a gep into a "bitcast" if
 the gep has more than one use.

Fixes PR10322.

llvm-svn: 134883
---
 .../InstCombine/InstructionCombining.cpp         |  8 ++++++++
 .../test/Transforms/InstCombine/getelementptr.ll | 16 ++++++++++++++++
 2 files changed, 24 insertions(+)

diff --git a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
index 92c10f5546c0..278016c6ece0 100644
--- a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -785,6 +785,14 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
   // getelementptr instructions into a single instruction.
   //
   if (GEPOperator *Src = dyn_cast<GEPOperator>(PtrOp)) {
+
+    // If this GEP has only 0 indices, it is the same pointer as
+    // Src. If Src is not a trivial GEP too, don't combine
+    // the indices.
+    if (GEP.hasAllZeroIndices() && !Src->hasAllZeroIndices() &&
+        !Src->hasOneUse())
+      return 0;
+
     // Note that if our source is a gep chain itself that we wait for that
     // chain to be resolved before we perform this transformation.  This
     // avoids us creating a TON of code in some cases.
diff --git a/llvm/test/Transforms/InstCombine/getelementptr.ll b/llvm/test/Transforms/InstCombine/getelementptr.ll
index aa2eba177b19..26c0e47f4b14 100644
--- a/llvm/test/Transforms/InstCombine/getelementptr.ll
+++ b/llvm/test/Transforms/InstCombine/getelementptr.ll
@@ -456,3 +456,19 @@ define i32* @test38(i32* %I, i32 %n) {
 ; CHECK: = sext i32 %n to i64
 ; CHECK: %A = getelementptr i32* %I, i64 %
 }
+
+; Test that we don't duplicate work when the second gep is a "bitcast".
+%pr10322_t = type { i8* }
+declare void @pr10322_f2(%pr10322_t*)
+declare void @pr10322_f3(i8**)
+define void @pr10322_f1(%pr10322_t* %foo) {
+entry:
+  %arrayidx8 = getelementptr inbounds %pr10322_t* %foo, i64 2
+  call void @pr10322_f2(%pr10322_t* %arrayidx8) nounwind
+  %tmp2 = getelementptr inbounds %pr10322_t* %arrayidx8, i64 0, i32 0
+  call void @pr10322_f3(i8** %tmp2) nounwind
+  ret void
+
+; CHECK: @pr10322_f1
+; CHECK: %tmp2 = getelementptr inbounds %pr10322_t* %arrayidx8, i64 0, i32 0
+}