From 0286ca89f0734a81c273e80732a573f619c7ace4 Mon Sep 17 00:00:00 2001
From: Nadav Rotem <nadav.rotem@intel.com>
Date: Tue, 5 Apr 2011 14:29:52 +0000
Subject: [PATCH] InstCombine optimizes gep(bitcast(x)) even when the bitcasts
 casts away address space info. We crash with an assert in this case. This
 change checks that the address space of the bitcasted pointer is the same as
 the gep ptr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@128884 91177308-0d34-0410-b5e6-96231b3b80d8
---
 .../InstCombine/InstructionCombining.cpp      | 19 +++++++++++--------
 test/Transforms/InstCombine/gep-addrspace.ll  | 16 ++++++++++++++++
 2 files changed, 27 insertions(+), 8 deletions(-)
 create mode 100644 test/Transforms/InstCombine/gep-addrspace.ll

diff --git a/lib/Transforms/InstCombine/InstructionCombining.cpp b/lib/Transforms/InstCombine/InstructionCombining.cpp
index 21851768cbb..4be671f031e 100644
--- a/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -849,22 +849,23 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
         GetElementPtrInst::Create(Src->getOperand(0), Indices.begin(),
                                   Indices.end(), GEP.getName());
   }
-  
+
   // Handle gep(bitcast x) and gep(gep x, 0, 0, 0).
   Value *StrippedPtr = PtrOp->stripPointerCasts();
-  if (StrippedPtr != PtrOp) {
-    const PointerType *StrippedPtrTy =cast<PointerType>(StrippedPtr->getType());
+  const PointerType *StrippedPtrTy =cast<PointerType>(StrippedPtr->getType());
+  if (StrippedPtr != PtrOp &&
+    StrippedPtrTy->getAddressSpace() == GEP.getPointerAddressSpace()) {
 
     bool HasZeroPointerIndex = false;
     if (ConstantInt *C = dyn_cast<ConstantInt>(GEP.getOperand(1)))
       HasZeroPointerIndex = C->isZero();
-    
+
     // Transform: GEP (bitcast [10 x i8]* X to [0 x i8]*), i32 0, ...
     // into     : GEP [10 x i8]* X, i32 0, ...
     //
     // Likewise, transform: GEP (bitcast i8* X to [0 x i8]*), i32 0, ...
     //           into     : GEP i8* X, ...
-    // 
+    //
     // This occurs when the program declares an array extern like "int X[];"
     if (HasZeroPointerIndex) {
       const PointerType *CPTy = cast<PointerType>(PtrOp->getType());
@@ -975,7 +976,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
       }
     }
   }
-  
+
   /// See if we can simplify:
   ///   X = bitcast A* to B*
   ///   Y = gep X, <...constant indices...>
@@ -983,12 +984,14 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
   /// analysis of unions.  If "A" is also a bitcast, wait for A/X to be merged.
   if (BitCastInst *BCI = dyn_cast<BitCastInst>(PtrOp)) {
     if (TD &&
-        !isa<BitCastInst>(BCI->getOperand(0)) && GEP.hasAllConstantIndices()) {
+        !isa<BitCastInst>(BCI->getOperand(0)) && GEP.hasAllConstantIndices() &&
+        StrippedPtrTy->getAddressSpace() == GEP.getPointerAddressSpace()) {
+
       // Determine how much the GEP moves the pointer.  We are guaranteed to get
       // a constant back from EmitGEPOffset.
       ConstantInt *OffsetV = cast<ConstantInt>(EmitGEPOffset(&GEP));
       int64_t Offset = OffsetV->getSExtValue();
-      
+
       // If this GEP instruction doesn't move the pointer, just replace the GEP
       // with a bitcast of the real input to the dest type.
       if (Offset == 0) {
diff --git a/test/Transforms/InstCombine/gep-addrspace.ll b/test/Transforms/InstCombine/gep-addrspace.ll
new file mode 100644
index 00000000000..4c971072cfb
--- /dev/null
+++ b/test/Transforms/InstCombine/gep-addrspace.ll
@@ -0,0 +1,16 @@
+; RUN: opt < %s -instcombine -S
+
+%myStruct = type { float, [3 x float], [4 x float], i32 }
+
+; make sure that we are not crashing when creating an illegal type
+define void @func(%myStruct addrspace(1)* nocapture %p) nounwind {
+ST:
+  %A = getelementptr inbounds %myStruct addrspace(1)* %p, i64 0
+  %B = bitcast %myStruct addrspace(1)* %A to %myStruct*
+  %C = getelementptr inbounds %myStruct* %B, i32 0, i32 1
+  %D = getelementptr inbounds [3 x float]* %C, i32 0, i32 2
+  %E = load float* %D, align 4
+  %F = fsub float %E, undef
+  ret void
+}
+